Epistemological positioning and evidentiality in English ...
A corpus-based analysis of evidentiality in popularised...
Transcript of A corpus-based analysis of evidentiality in popularised...
Ghent University
Faculty of Arts and Philosophy
Academic year 2012-2013
A corpus-based analysis of evidentiality in
popularised scientific discourse
Master paper submitted in partial fulfilment of the requirements for the degree of
“Master in de Taal- en Letterkunde: Engels – Spaans”
Annelien Siau
Supervisor: Dr. Julie Van Bogaert
2
3
Ghent University
Faculty of Arts and Philosophy
Academic year 2012-2013
A corpus-based analysis of evidentiality in
popularised scientific discourse
Master paper submitted in partial fulfilment of the requirements for the degree of
“Master in de Taal- en Letterkunde: Engels – Spaans”
Annelien Siau
Supervisor: Dr. Julie Van Bogaert
4
Acknowledgements
I would like to thank the following people for their support during the composition of this
dissertation.
First and foremost, I want to thank my supervisor Dr. Julie Van Bogaert. She was
always there to answer my numerous questions, provided helpful remarks and made time
to correct my writing.
I would also like to thank Mien Stoffels, a great friend who supported me from
beginning to end. I am very grateful for her encouraging words and useful advice.
Finally, a great thank you goes to my family and friends, who always supported me
and never lost their confidence in me.
5
Table of contents
ACKNOWLEDGEMENTS .................................................................................................... 4
LIST OF FIGURES .............................................................................................................. 7
LIST OF TABLES ............................................................................................................... 8
INTRODUCTION ............................................................................................................... 9
1 EVIDENTIALITY ...................................................................................................... 12
1.1 EVIDENTIALITY IN NON-EUROPEAN LANGUAGES ........................................................................... 12
1.1.1 Defining the concept: evidentiality and modality ...................................................................... 13
1.1.2 Types of evidence .............................................................................................................................. 16
1.2 EVIDENTIALITY IN ENGLISH ......................................................................................................... 18
1.2.1 Discussion ............................................................................................................................................ 18
1.2.2 Types .................................................................................................................................................... 20
1.3 EVIDENTIALITY IN DUTCH ........................................................................................................... 23
1.4 IMPORTANCE OF EVIDENTIALITY .................................................................................................. 24
1.4.1 Functional motivations for evidentiality ..................................................................................... 25
1.4.2 Evidentiality in scientific discourse .............................................................................................. 26
2 AUDIOVISUAL TRANSLATION ................................................................................ 28
2.1 SUBTITLING ................................................................................................................................ 29
2.1.1 Definition ............................................................................................................................................. 29
2.1.2 Strategies ............................................................................................................................................. 30
3 METHODOLOGY ...................................................................................................... 34
3.1 DATA COLLECTION ...................................................................................................................... 34
3.2 CRITERIA .................................................................................................................................... 35
6
3.3 METHODS ................................................................................................................................... 37
4 RESULTS ................................................................................................................. 39
4.1 CLASSIFICATION AND INVESTIGATION OF THE ENGLISH DATA ......................................................... 39
4.1.1 The English evidentiality types .................................................................................................... 39
4.1.1.1 Direct evidence ..................................................................................................... 40
4.1.1.2 Reported evidence ............................................................................................... 41
4.1.1.3 Inferential evidence ............................................................................................ 44
4.1.1.4 Expression of (un)certainty ................................................................................ 46
4.1.2 Formal encoding ................................................................................................................................ 48
4.1.3 Functional motivations for evidentiality ..................................................................................... 50
4.1.3.1 Engagement positions ......................................................................................... 51
4.1.3.2 Summary and interpretation ............................................................................. 55
4.2 ENGLISH SPEECH VERSUS DUTCH SUBTITLES ................................................................................. 58
4.2.1 Differences .......................................................................................................................................... 58
4.2.1.1 Types ...................................................................................................................... 59
4.2.1.2 Formal encoding in Dutch .................................................................................. 63
4.2.2 Influence of subtitles......................................................................................................................... 68
4.2.3 Functional motivations of the subtitler ........................................................................................ 73
5 DISCUSSION ........................................................................................................... 76
5.1 TYPES OF EVIDENTIALITY IN POPULARISED SCIENCE ....................................................................... 76
5.2 THE FORMAL ENCODING OF EVIDENTIALITY IN POPULARISED SCIENCE ............................................. 78
5.3 THE FUNCTIONAL MOTIVATIONS FOR EVIDENTIALITY IN POPULARISED SCIENCE ............................... 79
6 CONCLUSION .......................................................................................................... 81
7 REFERENCES ........................................................................................................... 83
7
List of figures
FIGURE 1. COMBINATION OF THE TYPES OF EVIDENCE IN NON-EUROPEAN LANGUAGES ............................... 18
FIGURE 2. LEXICAL EXPRESSIONS OF ENGLISH EVIDENTIALITY ................................................................... 22
FIGURE 3. RESPONSIBILITY FOR THE STATEMENT EXPRESSED BY THE SPEAKER ........................................... 56
FIGURE 4. FROM CONTRACTION TO EXPANSION ........................................................................................ 56
8
List of tables
TABLE 1. INCIDENCE OF TYPES OF EVIDENCE ............................................................................................. 39
TABLE 2. FREQUENCY OF THE STRUCTURES NP + VERB, PN + VERB AND PRON + VERB ............................... 49
TABLE 3. CHANGE OF TYPES FROM ENGLISH TO DUTCH ............................................................................. 60
TABLE 4. TIME ANALYSIS OF A 'DELETION' EXAMPLE ................................................................................. 70
TABLE 5. TIME ANALYSIS OF A 'DELETION' EXAMPLE ................................................................................. 71
9
Introduction
Evidentiality is a linguistic phenomenon acknowledged in the 20th century. The term
appears for the first time in Boas' posthumously published Kwakiutl Grammar (1947), in
which he recognises "a small group of suffixes [which] expresses source and certainty of
knowledge" (Boas et al. 1947 qtd. in Jacobsen 1986: 4). He uses the term 'evidential' to refer
to one of the suffixes. Boas' students (Edward Sapir, Morris Swadesh, and Harry Hoijer) also
recognise the concept of evidentiality in their works on Indian languages, but it was a
friend of Boas, Roman Jakobson, who gave evidentiality its basic definition. Jakobson (et al.
1984: 46) defines 'evidential' in 1957 as "a tentative label for the verbal category which takes
into account three events - a narrated event, a speech event and a narrated speech event
(Ens), namely the alleged source of information about the narrated event." Furthermore, he
suggests four possible sources of evidential information: someone else's report (quotative,
i.e. hearsay evidence), a dream (relative evidence), a guess (presumptive evidence) or one's
own previous experience (memory evidence) (Jakobson et al. 1984). This definition, which
soon became widely spread, introduced the concept of evidentiality in the field of
linguistics. Twenty-four years after Jakobson's publication, in 1981, several linguists
gathered at a symposium in Berkeley with the intention
to compare evidentiality in a variety of languages and to explore such
general questions as the areas of epistemology for which different
languages provide evidential markings, the nature of such markings, and
the ways in which they arise and spread.
(Chafe & Nichols 1986: viii)
This conference resulted in an edited volume entitled Evidentiality: the Linguistic Coding of
Epistemology (Chafe & Nichols 1986). This collection of articles firmly established the notion
of evidentiality in linguistic circles, and soon put it at the centre of various discussions.
Starting out as the topic of typological studies, evidentiality spread to grammaticalisation,
cognitive linguistics, syntax and pragmatics (Dendale & Tasmowski 2001). In addition, with
scholars increasingly trying to identify the evidential systems of various languages, it
spread from American Indian to European languages as well.
Following the example of those scholars, this dissertation deals with evidentiality in
English. This subject is of particular interest and complexity because of the disagreement
on how to define the concept in languages – like English – lacking grammatical evidentiality
10
markers. In fact, some scholars deny that evidentiality exists at all in such languages. To
refute this statement, this dissertation examines how evidentiality can be expressed in
English. The delineation of a possible English evidentiality system is done in the literature
review. Subsequently, I investigate the features of the English evidentiality used in a
popularised form of science: the television documentary. The corpus consists of English
evidentiality expressions and their concomitant Dutch subtitles, taken from the first season
of the American science documentary How the Earth Was Made. This corpus-based approach
is adopted to examine the effect of the specific genre on the evidential forms and types.
Additionally, to allow for an adequate amount of results, the Dutch subtitles of the
evidentiality expressions are studied as well. This comparative interlingual study serves to
gain more insight into the multifunctional meaning of evidential expressions.
Since the dissertation tackles two notions, i.e. the English evidentiality expressions
and their Dutch subtitles, the theoretical framework provided in the literature review
equally consists of two parts. First, the sources about evidentiality are compared and
discussed, with the aim to establish a workable definition to rely on when constituting the
corpus and investigating the data. Subsequently, evidentiality in English and in Dutch are
explored, followed by a section about the functional motivations for evidentiality. In
particular, the section about English evidentiality is based on the evidential forms provided
by Ifantidou (2001), while the sections about Dutch evidentiality and functional motivations
are inspired by one of the few scholars who have studied evidentiality in Dutch, Ferdinand
de Haan (2000). In the second part, the main subtitling techniques used in an interlingual
medium are discussed, as it is important to be aware of how the multimodal genre may
influence the translation of the evidentiality expressions.
The literature review is followed by the results of the corpus analysis, which again
utilised the twofold structure of the theoretical introduction. The first study defines
evidentiality for the particular English data. The first section of this study classifies the data
into the four evidentiality types proposed in the literature review (i.e. the indication of
direct evidence, reported evidence, inferential evidence and the expression of
(un)certainty), and investigates their specific characteristics. The second section focuses on
the formal encoding of the evidential data, while the third examines the functional
motivations of the speaker.
Following the investigation of the English evidential data, the second study is
concerned with the subtitles. These may provide additional information about the
11
meanings of English evidentials when used in scientific discourse. The interlingual
comparative analysis of the subtitles puts emphasis on the differences between the source
sentences and the subtitles.
Finally, I discuss how these differences come into existence, and whether or not they
are the consequence of a different functional motivation. By providing an answer to these
questions, this study doubles as a study on subtitling and the ways in which subtitles may
affect the English evidentiality expressions.
12
I. LITERATURE REVIEW
1 Evidentiality
The current part tackles the notion of evidentiality in four sections. The primary concern of
this part is the delineation of the domain of evidentiality for this particular dissertation.
First of all, to get acquainted with the concept of evidentiality, section 1.1 presents
evidentiality in non-European languages by outlining various definitions and the most
generally acknowledged types. The findings of section 1.1 are used as a basis in order to
create a suitable evidentiality system for English (section 1.2) and for Dutch (section 1.3).
These systems are used as a starting point for the composition of the corpus and the
analysis of the English evidentiality data. Finally, in the last part, by discussing the
functional motivations for evidentiality, I emphasise the significance of evidentiality in
general and in particular in scientific discourse.
1.1 Evidentiality in non-European languages
As mentioned in the introduction, the first acknowledgment of an evidential structure was
done in a study about Kwakiutl. The fact that evidentiality was first brought to light in
studies of American Indian languages is not surprising given that in those languages "the
marking of evidentiality through verb suffixes is widespread" (Chafe & Nichols 1986: viii).
This observation explains why the first studies on evidentiality deal with evidentiality as a
grammatical phenomenon and focus on languages spoken in various parts of North and
South America (cf. Jacobsen on Makah; Oswalt on Kashaya; Schlichter on Wintu; Whistler on
Patwin; Gordon on Maricopa). Defining evidentiality for European languages, which
generally lack evidential verb suffixes, is a difficult task. In order to arrive at an appropriate
overview of English evidentiality markers and a functional definition of English
evidentiality, I begin from the starting point: the grammatical evidentiality in non-
European languages.
13
1.1.1 Defining the concept: evidentiality and modality
One of the pioneers of the investigation of evidentiality as a grammatical category is
Alexandra Y. Aikhenvald. In her book Evidentiality (Aikhenvald 2004), she gives the following
definition:
In about a quarter of the world's languages, every statement must
specify the type of source on which it is based - for example, whether the
speaker saw it, or heard it, or inferred it from indirect evidence, or
learnt it from someone else. This grammatical category, whose primary
meaning is information source, is called 'evidentiality'.
(Aikhenvald 2004: 1)
This definition deals with the conceptual notion of evidentiality. Aikhenvald states that the
main meaning of evidentials is the indication of the source of information on which a
statement is based. This is possible via visual, auditive, inferred, or hearsay evidence.
Similarly, in the oft-cited definition by Anderson (1986: 274) it is said that evidentials give
"the kind of justification for a factual claim which is available to the person making that
claim" and this indication of evidence has to be the primary meaning of the evidential
structure. Nevertheless, not all linguists agree with this one and only meaning given to
evidentiality.
On the one hand there exists a consensus that "[t]he basic characteristic of linguistic
evidentiality is the explicit encoding of a source of information or knowledge (i.e. evidence)
which the speaker claims to have made use of for producing the primary proposition of the
utterance" (Diewald & Smirnova 2010: 1). According to Willett (1988), this view corresponds
to evidentiality in the narrow sense, because an explicit relationship between evidentiality
and modality is denied. On the other hand, there are various opinions regarding
evidentiality in the broad sense. For instance, one can deduce from the earliest explanation
of the concept of evidentiality by Boas, that evidentiality not only refers to source of
knowledge, but also to certainty of knowledge. This expression of (un)certainty is generally
denoted as epistemic modality. According to Dendale and Tasmowski (2001: 343), "the forms
marking the source of information also mark the speaker's attitude", which explains the
difficulty with the interface between evidentiality and modality when trying to define the
concept of evidentiality.
14
As Dendale and Tasmowski (2001: 341) note, one of the main difficulties raised by
researchers concerning the conceptual problem of the term evidentiality is "the question of
the scope and definition of the terms evidentiality and evidential and their relation to the
terms epistemic modality and epistemic modal marker." To answer this question they suggest
three relations between evidentiality and modality, being disjunction (where they are
conceptually distinguished from each other), inclusion (where one is regarded as falling
within the semantic scope of the other), and overlap (where they partly intersect) (Dendale
& Tasmowski 2001). Disjunction is considered as a characteristic of evidentiality in the
narrow sense, while inclusion and overlap refer to a broad definition. As noted above,
Aikhenvald (2004: 5) argues for a relation of disjunction, denying "any reference to validity
or reliability of knowledge or information." In addition, according to a more recent study,
"there has been growing acceptance of the assumption that evidentiality is a semantic-
functional domain in its own right, and not a sub-division of epistemic modality" (Diewald
& Smirnova 2010: 2). This narrow view is typical of linguists who look at languages which
encode evidential information in their inflectional morphology (Ifantidou 2001). In other
words, they investigate evidentiality as a grammatical phenomenon. By contrast, according
to Willett (1988), the relation between evidentiality and modality is one of inclusion. He
clarifies this by saying "that evidential distinctions are part of the marking of epistemic
modality, even though evidentials as such are seldom explicitly mentioned in theoretical
treatments of modality" (Willett 1988: 52).
Other broad views on the concept of evidentiality are proposed by Chafe and
Mithun. Chafe (1986), as one can see in the title of the Berkeley edition Evidentiality:
Linguistic Coding of Epistemology, considers evidentiality as an equivalent for epistemology.
He includes everything that involves attitudes towards knowledge. The diagram below
explains his approach.
Source of knowledge Mode of knowing Knowledge matched against
Reliable
??? --> Belief |
Knowledge
|
|
Evidence --> Induction --> verbal resources
Language --> Hearsay --> expectations
Hypothesis --> Deduction
Unreliable
(Chafe 1986: 263)
15
In Chafe's analysis, evidentiality refers not only to the source of knowledge and the mode of
knowing but also to the reliability of that knowledge and that knowledge matched against
verbal resources and expectations. Thus, Chafe regards evidentiality as "an indication of the
source and reliability of a speaker's knowledge" (Willett 1988: 55). Mithun (1986) does the
same. She indicates three ways to qualify the reliability of an utterance: specify the degree
of precision or truth, specify the probability of its truth, or specify expectations concerning
their probability (Mithun 1986). These broad views typically include both grammatical and
lexical encoding of evidentiality.
Finally, when modality and evidentiality partly intersect, one speaks of an
overlapping relation. The interface is then expressed by evidential inference, "which [van
der Auwera and Plungian] claim to be identical to the modal value of epistemic necessity"
(Dendale & Tasmowksi 2001: 342).
This dissertation applies the concept of evidentiality in its broadest sense because of
the following reasons. First of all, in order to create an appropriate evidentiality system for
languages that lack grammatical evidentiality markers, it is easier to start from a broad
point of view since in those languages "the reason for indicating a source of information is
often to give the interlocutor an idea of the degree of certainty or reliability of a given piece
of information (Frajzyngier 1985; Willett 1988; Dendale 1991; Bybee et al. 1994; Boye &
Harder 2009)" (Van Bogaert & Dendale 2013: 4). When dealing with English, for instance, it
is necessary to include non-grammatical forms of evidentiality due to the fact that it hardly
has any grammaticalised evidential markers.1 Secondly, in the light of this specific study, it
is recommended to include the epistemic attitude of the speaker. The necessity of this
inclusion of (un)certainty is further explained in section 1.4, where the importance of
evidentials in a scientific discourse is discussed. Connected to this is the opinion of Van
Bogaert and Dendale, who argue for a relativistic approach to the definition of evidentiality:
For us, how one delineates evidentiality and whether one can assign
evidential status to a given linguistic expression and designate it with
the term evidential (instead of something like information source marker) is
contingent on the aim of the study at hand and on the research
paradigm in which it inscribes itself.
(Van Bogaert & Dendale 2013: 24)
1 For a discussion on must see section 1.2.1.
16
One of the aims of this study is to investigate the occurrence of English evidentiality in a
scientific genre, namely television documentary. Consequently, it is not useful to adopt the
strict definition of Aikhenvald, who is interested in typological studies of grammatical
evidentiality. As will be shown in sections 1.2, 1.3 and 1.4, this study is more in line with
discourse analyses and defines evidentiality in a functional light. Therefore, "it seems
perfectly reasonable to incorporate a large variety of expressions, whatever status they
have, whether grammatical or lexical and to include pragmatically inferred meanings" (Van
Bogaert & Dendale 2013: 24).
1.1.2 Types of evidence
The indication of the information source, being the basic semantic function of evidentials,
can be expressed in various ways. Scholars discovered different categories while studying
American Indian languages. A vital and frequently cited example of such a study is Barnes’s
(1984) work on Tuyuca. Barnes distinguishes five evidentiality types based on how the
information is received. Below, examples are given for each evidential category, viz. (1)
visual evidence, (2) nonvisual evidence (to indicate any of the senses other than visual), (3)
apparent evidence, (4) secondhand evidence and (5) assumed evidence.
(1) diiga ape-wi
'He played soccer'. (I saw him play.)
(2) diiga ape-ti
'He played soccer'. (I heard the game and him, but I didn't see it or him.)
(3) diiga ape-yi
'He played soccer'. (I have seen evidence that he played: his distinctive
shoeprint on the playing field. But I did not see him play.)
(4) diiga ape-yigi
'He played soccer'. (I obtained the information from someone else.)
(5) diiga ape-hiyi
'He played soccer'. (It is reasonable to assume that he did.)
(Barnes 1984: 257)
Another well-known classification based on the study of several American Indian languages
is Willett’s (1988). The main parameter of his categorisation is direct versus indirect
evidence. Willett speaks of direct or attested evidence when the speaker was a direct
17
witness to the source of information. Attested evidence can be obtained by the visual sense,
the auditory sense or one of the other three senses. Indirect evidence implies that the
source of the speaker's information is of secondary nature and includes reported (evidence
via verbal report) and inferring (based on inference) evidence. Reported evidence can be
second-hand or third-hand. The former denotes hearsay information received from a direct
witness, while the latter represents hearsay information passed on from one person to
another and finally to the speaker. The third reported evidence option, namely folklore,
refers to information that is part of the oral literature, like myths and legends. Finally, the
inferring evidence may involve results, i.e. observable evidence, or a mental construct,
named reasoning. (Willett 1988)
A more recent model, partly based on Willett, is the one proposed by Aikhenvald:
1. VISUAL (SENSORY): information acquired through seeing
2. (NON-VISUAL) SENSORY: information acquired through other forms of
sensory perception (smell, taste, touch)
3. INFERENCE: conclusion based on visual or tangible evidence
4. ASSUMPTION: based on indications other than visible evidence (logic,
supposition, general knowledge), with a strong reasoning component
5. HEARSAY: reported information without making reference to the
person from whom the information was acquired
6. QUOTATIVE: reported information with explicit mention of the source
(Aikhenvald 2004: 63-64)
Categories 1 and 2 coincide with the direct evidence category of Willett, even though they
lack the explicit mention of the auditory evidence. The others are part of indirect evidence.
Aikhenvald's term ‘assumption’ is comparable to Willett's ‘reasoning’. Among the
differences are the exclusion of the first-hand versus second-hand hearsay distinction, the
deletion of the notion folklore, and the addition of a second type of reported evidence, i.e.
quotative. The term quotative is also mentioned by Anderson (1986). In the general
category of reported evidence he distinguishes at least four subdivisions: "(a) hearsay, (b)
general reputation, (c) myth and history (these three being evidentials), and (d) 'quotative'
(marginally an evidential)" (Anderson 1986: 289). Even though he calls the quotative
evidence marginally an evidential, he follows Aikhenvald by saying that it stands for "This
is what X said" (Anderson 1986: 289). Other linguists use different terms. Chafe (1986: 269)
18
calls all reported evidence hearsay evidence and states that the construction used to cite a
reference, i.e. quotative, is a "hearsay evidential expressed in [its] most precise and
deliberate form." Regarding the reported evidence, I will use the more general terms
defined by Aikhenvald: hearsay and quotative. The distinction is made between mentioning
or not mentioning of the source, which means that these categories can also comprise the
subdivisions made by Willett and Anderson. For instance, the sentence The Romans believed
that Romulus and Remus founded Rome, refers to a myth and is analysed as indicating
quotative evidence, while Romulus and Remus are believed to have founded Rome denotes the
same myth but refers to hearsay.
As a summary, and to conclude this section of evidential subcategories based on
American Indian languages, the three taxonomies are combined in Figure 1.
Types of
evidence
Direct
Attested VISUAL
NON-VISUAL
Indirect
Reported HEARSAY
QUOTATIVE
Inferring INFERENCE
ASSUMPTION
FIGURE 1. Combination of the types of evidence in non-European languages
1.2 Evidentiality in English
1.2.1 Discussion
In the majority of the studies on evidentiality it has been assumed that European languages
lack "grammatical markers and grammatical systems of evidentiality" (Diewald & Smirnova
2010: 2). Aikhenvald (2004) addresses this problem for English by saying that English does
not have pure evidential markers and therefore compares evidentiality with other
categories, like modality. She also draws a distinction between pure evidential markers,
which are grammatical, and evidential strategies, which are lexical or pragmatic, and
concludes that in most European languages the evidential structures are 'mere' evidential
strategies. Even though Aikhenvald states that some languages lack a specific evidential
grammatical category, she does admit that referring to a source of information is universal.
19
A lot of linguists agree with this statement and apply it to English, as can be seen in the
following overview of citations:
Although English does not have a specific grammatical category of
evidentials, a variety of optional, nonpropositional constructions can
function as evidentials.
(Barton 1993: 746)
The definition of evidentials in English thus has to be a functional one.
(Barton 1993: 746)
English has a rich repertoire of evidential devices. It expresses
evidentiality with modal auxiliaries, adverbs, and miscellaneous
idiomatic phrases, although not, for example, with a coherent set of verb
suffixes like those in some California Indian languages.
(Chafe 1986: 261)
One undercover evidential in English is the inferential value of
polysemous must, distinct from its obligational one, as nicely delineated
by Chafe (1970:179-84).
(Jacobsen 1986: 7)
There is general agreement as to the notion that English does not have a well-defined
grammatical system to express evidentiality. Therefore it is advisable to define English
evidentiality in a functional light, by using lexical means. However, saying that English does
not at all have grammatical evidential markers needs some consideration since there is one
borderline case; the modal must.
The criteria to distinguish grammatical evidentials from lexical evidentials are not
clear-cut. According to de Haan (2000), a grammaticalised evidential morpheme is
characterised by the following four criteria:
1. Evidentials are not themselves the main part of the clause.
2. Evidentials do not show agreement with the speaker.
3. The morphemes have the expression of evidentiality as their primary
meaning.
4. Evidentials cannot be in the scope of a negative element.
(de Haan 2000: 75-76)
20
The first and second criteria are responsible for the exclusion of sentences like it is evident
that and I see that (de Haan 2000). The fourth criterion is the reason why de Haan considers
the Dutch verb moeten as an evidential (cf. section 1.3), whereas the third contributes to the
elimination of must from the evidential system. Criterion 3 "is used to distinguish between
true evidentials and those elements for which evidentiality is only inferentially present" (de
Haan 2000: 75). To illustrate these inferential evidentials, Anderson (1986: 275), who also
mentions the difference, gives four examples: (1) the toast is burnt, (2) the toast burned, (3) the
toast has burned, (4) the toast must have burned. De Haan (2000) says that from the third
statement the toast has burned, the second the toast burned may be inferred. In that case "the
perfect denotes, secondarily, the evidential notion of 'evidence for an action'", but cannot
be considered an evidential itself (de Haan 2000: 76). According to de Haan (2000), the modal
must is a similar example. Anderson (1986: 275), by contrast, considers the fourth example
as a true evidential because, "[w]hen the present state is used as circumstantial evidence for
inferring an unwitnessed past event, English normally adds the epistemic must of logical
inference." Consequently, must indicates that there is evidence for the action expressed, for
instance the smell of the burnt toast. De Haan argues that this is not enough proof for the
verb must to be treated as a grammaticalised evidential, since the expression of
evidentiality is not its primary meaning. Following Anderson, this dissertation proposes to
treat the epistemic modal must as a grammatical inferential evidential.
In addition, Ifantidou (2001) notices that sometimes a structure may function as an
evidential without this information being linguistically encoded, which is called pragmatic
inference. In this view, the sentence John is feeling miserable today, can be based on
observation (the speaker has seen his miserable expression), hearsay (the speaker repeats
what John told him), or inference (the speaker deduces the information from John's
behaviour) (Ifantidou 2001). (Ifantidou 2001)
1.2.2 Types
A good overview of the lexical structures that express English evidentiality is given by
Ifantidou (2001). Considering the definition of an evidential as a functional one, Ifantidou
organises the categories around the two main functions (in the broad sense). The first
function, indication of source of knowledge, can be obtained by observation
(sensory/perceptual evidence), by hearsay (from other people), by inference and by
memory. The observational evidence is mainly expressed by perception verbs like I see, I
21
hear, I feel, it tastes, or by verbs which express less reliability like looks like, sounds like, feels
like, smells like. Hearsay can be indicated by the expressions tells me, I hear, people say, he is
said, he is reputed, allegedly, reportedly. Ifantidou also includes less direct hearsay devices
which primarily perform other functions like it seems, it's supposed to, apparently. The
structures presumably, seems to/must be, must have, I gather are typical of inferential evidence
and even though not frequently treated as evidentials, Ifantidou adds so, I deduce, and
consequently. Finally, the information source can also be one's own memory, expressed by I
remember, I recall, as I recollect. (Ifantidou 2001: 5-7)
Secondly, Ifantidou's evidentiality system for English includes the speaker's degree
of certainty, characterised by propositional attitude and parenthetical expressions (I think,
I know, I suspect, I guess, I suppose), adverbials (probably, certainly, possibly, undoubtedly, surely,
evidently, obviously) and epistemic modals (may, might, can, could, must, will, ought to/should).
(Ifantidou 2001: 5-7)
The above-mentioned categorisation represents a broad view on evidentiality, based
on the suggestion that "[i]n its broadest sense, an utterance has an evidential function if
and only if it overtly communicates evidential information, whether this information is
linguistically encoded or pragmatically inferred" (Ifantidou 2001: 161). This broad view is
suggested by the inclusion of expressions which are not standardly treated as evidential.
These are lexical expressions not included by the majority of the scholars, but for Ifantidou
important enough to mention. Ifantidou adds and at the same time deletes some terms
previously seen. On the one hand, she takes into account the definition by Jakobson, who
includes memory as a possible evidential. She states that "[s]ince memory is variably
reliable, such expressions have a claim to be considered as evidentials" (Ifantidou 2001: 7).
On the other hand, she leaves out Aikhenvald’s ‘quotative’ and ‘assumption’ categories. As
already explained, this study will adopt Aikhenvald's terms for reported evidence.
To conclude, I give a scheme of the evidentiality types seen in section 1.1.2 (cf.
Figure 1) combined with the English lexical expressions proposed by Ifantidou. The
overview represented in Figure 2 serves as a basis for the composition of the corpus and the
English evidentiality discussed in the corpus analysis.
22
A. INDICATION OF SOURCE OF KNOWLEDGE
1. Direct evidence
a. Visual
I see, looks like
I see him swimming.
b. Non-visual I hear, I feel, it tastes, sounds like,
feels like, smells like.
I hear that the train is coming.
c. Memory I remember, I recall, as I recollect I remember that he was an
excellent professor.
2. Indirect evidence
a. Reported Hearsay I hear, he is said, he is reputed,
allegedly, reportedly, it seems, it's
supposed to, apparently
He is said to have committed
the crime.
Quotative X tells me, people say, they suggest Tom tells me John is the
burglar.
b. Inferring Inference Presumably, seems to,must
be/must have, I gather, guess,
suppose, so, I deduce, consequently
There is a car on the driveway.
I gather that Tom is in town
Assumption Must, I assume I assume that Tom is on
holiday.
B. SPEAKER'S DEGREE OF CERTAINTY
a. Propositional attitude and
parenthetical expressions
I think, I know, I suspect, I guess, I
suppose
I think he is a very clever
person.
b. Adverbials probably, certainly, possibly,
undoubtedly, surely, evidently,
obviously.
Obviously he did not kill the
cat.
c. Epistemic modals may, might, can, could, must, will,
ought to/should
It may be possible that he ran
away.
FIGURE 2. Lexical expressions of English evidentiality
23
1.3 Evidentiality in Dutch
In order to be able to analyse the Dutch subtitles of the English evidentiality expressions, a
concise overview of evidentiality in Dutch is appropriate. Nevertheless, studies about a
possible evidentiality system in Dutch are rather scarce. De Haan is one of the few linguists
who examine the properties of evidentiality in Dutch and presents five forms:
a. Modal verb moeten 'must'
De film moet uitstekend zijn
'The film is said to be excellent'
b. Past tense modal verb zou(den) 'should'
Bij de brand zouden alle bewoners zijn omgekomen
'All inhabitants are said to have perished in the fire'
c. Raising verb schijnen 'seem'
Jan schijnt ziek te zijn
'John seems to be ill'
d. Complements of perception verbs
Ik hoor, dat Jan ziek is
'I hear that John is ill'
e. "Quotative"
"Douane controleert El Al niet"
'Customs [officers] do not check El Al [planes]'
(de Haan 2000: 74)
Dividing these Dutch expressions into the evidential categories seen above is not an easy
task. First of all, the modal verbs moeten and zouden express either modality or evidentiality
(see below). With their evidential meaning they can have both a quotative and an
inferential interpretation. For instance, sentence a. can be interpreted as My friends say it is a
good film or I like the director's work and therefore this film must be good as well (de Haan 2001:
214). The raising verb schijnen is also marked by ambiguity. Example c. Jan schijnt ziek te zijn
can refer to a quotative, People say that John is ill, or an inferential, I infer from his absence that
he is ill. The perception verbs in Dutch refer to indirect evidence or in the case of example d.
to hearsay evidence (de Haan 2000). Finally, de Haan uses 'quotative' to denote the literal
citation of someone's words indicated by means of quotation marks. In the corpus analysis I
24
will examine whether these expressions are (frequently) used in the translation of the
English evidential constructions.
While de Haan (2000) excludes the English epistemic modal must from the
grammaticalised evidential system, he opts to treat the Dutch modal moeten as a true
evidential. He adopts this approach because of the agreement with criterion (4) Evidentials
cannot be in the scope of a negative element (cf. section 1.2.1). When the verb moeten co-occurs
with a negative element, moeten has scope over the negation (de Haan 2000). This is only
possible when dealing with the evidential interpretation of moeten. If moeten is replaced by
another modal verb (hoeven, kunnen) the evidential interpretation gets lost and the modal
verb is in the scope of the negative element (de Haan 2000). Secondly, the other criteria also
hold (cf. section 1.2.1). The verb moeten is not the main part of the clause nor does it show
agreement with the speaker. In isolation the verb moeten accepts different interpretations.
As illustration, sentence a. De film moet uitstekend zijn can have a deontic reading (it is
required to be a good film), an epistemic interpretation (it is probable that it is a good film), and
an evidential interpretation (there is evidence that it is a good film) (de Haan 2000). However,
de Haan argues that, seen in the context, the main interpretation of the verb moeten is the
evidential one. It has indication of evidence as its primary meaning, which coincides with
criterion 3. (de Haan 2000)
The three-way ambiguity of moeten is caused by grammaticalisation. Diachronically,
epistemic modality (Stage II), which expresses probability based on evidence, is developed
from deontic modality (Stage I). Evidential moeten (Stage III), which expresses only
evidence, arose from epistemic moeten by grammaticalisation (de Haan 2000). De Haan
(2000) claims that the English epistemic modal must has not (yet) arrived at the third stage
and therefore cannot be considered as a true evidential. As outlined in section 1.2.1, in this
dissertation I will not agree with de Haan, because I consider There is evidence that the toast
has burned; the house is full of smoke as to have an evidential value comparable to The toast
must have burned.
1.4 Importance of evidentiality
The classification of the English evidentiality system (cf. Figure 2) is based on the main
meanings of evidentiality: expressing the source of knowledge and the certainty of
knowledge. Inside this division several functional motivations for evidentiality can be
found, especially in scientific discourse.
25
1.4.1 Functional motivations for evidentiality
Considering direct evidence, the speaker is a witness of the actions described and therefore
makes "him/herself into a reference point" (de Haan 2001: 217). According to de Haan
(2001), receiving the information from a first-hand source implies high reliability and
responsibility. Reportives, on the other hand, remove the responsibility from the speaker.
Because of the introduction of another person, the speaker outsources the responsibility for
the information expressed, which has two consequences. On the one hand, by
acknowledging that the statement is not witnessed by the speaker, he/she disavows
responsibility for a claim he/she does not agree with or is not certain of. In English this may
be expressed by it is said, people think, etc. On the other hand, it is a useful way to assign
authority to the statement. By adding a second convincing party, like experts in the field,
the statement can be considered more valuable and truthful. (de Haan 2001)
Another way to reinforce this authority is to substantiate the argument with clear
evidence. For instance in scientific or academic writing, it is appropriate to ask how one has
obtained particular information or whether there exists proof for a certain statement.
These questions can be answered by reportive markers, as explained above, or by
inferential evidentials. De Haan (2001: 193) defines inference as "the grammaticalized way
of showing that the speaker makes his/her statement based on a deduction from facts, and
not on a direct observation of the action itself." Examples of these facts can be footprints in
the snow indicating the presence of a human or animal, or bread crumbs as the evidence of
a loaf of bread which once existed. The statement can be inferred from the available
evidence (1a) or the evidence can be expressed overtly (1b):
(1a) bóahõã-yu
rot-INFER:OTHER:PAST
'It rotted' (said of a plant after pulling it up to examine it)
(1b) Dáman dara-t vac dyi turasno.
shallow sit-PI INFER ART peach
Pui' cu-t va' ia'ray gu jivihl.
Thus so-PI then. fell ART wind
'These peach trees must have been planted shallowly. That's why the wind
blew them over'.
(de Haan 2001: 193)
26
Thus, the significance of evidentials lays firstly, in the fact that they permit the speaker to
acknowledge or deny responsibility for a statement, and secondly, in the possibility to
assign authority to the statement in order to make it more trustworthy. The latter can be
realised by adding a second persuasive information source, or evidence from which the
information is inferred. Additionally, dealing with evidentiality in English, I have included
the function of expressing (un)certainty towards an utterance via epistemic modals,
adverbials or parenthetical expressions. In the following section I try to prove that these
functional motivations are also highly valuable in scientific writing.
1.4.2 Evidentiality in scientific discourse
Ahmad (2012) examines the stylistic features of scientific English and distinguishes it from
literary language because of the following features:
The scientific language is accurate, precise, and detached from individual
impulse. It aims to inform about an important issue and what particular
approach is taken up to investigate that issue. It is an objective
interpretation of facts and findings. It contains such components and
findings that need external and experimental evidence to consolidate
their validity. (...) The pursuit of universal generalization in scientific
texts enables the author to signal credibility, reliability, objectivity and
ultimately authority to their readers and the research community.
(Ahmad 2012: 47, emphasis added by author)
On the one hand, the citation above is significant because it defines scientific English as
detached from individual impulse and objective, which is a statement I discuss in the second
part of this section. On the other hand, it emphasises the need to provide evidence, which
can be linked to the main role of evidentiality (information source), the possibility to signal
(un)certainty, which is the second function of English evidentiality (cf. Figure 2) and the
assignability of authority, one of the motivations outlined in the previous section. In spite
of these observations which stress the evaluative and informative nature of scientific
writing, little is written about the application of evidential markers in a scientific discourse.
Consequently, this part deals with a concept closely related to evidentiality, i.e. hedging.
Hedging, "the expression of tentativeness and possibility" (Ahmad 2012: 52), is one of the
features of scientific English delineated by Ahmad. Ahmad (2012: 52) incorporates in the
27
notion of hedging, linguistic devices such as epistemic main verbs (to indicate, to suggest, to
propose, to tend, to seem, to appear), epistemic modal auxiliaries (may, might, can, could) and
epistemic adverbials like hypothetically, probably, likely. Accordingly, from the overlap
between the epistemic hedging expressions mentioned by Ahmad and the English
evidential expressions categorised above (cf. Figure 2), one can deduce that evidentiality
also is a vital feature of scientific writing.
In view of the close connection between hedging and evidentiality, I refer to another
source in which Loredana Frățilă (2007) associates hedging with objectivity. As claimed by
Ahmad, scientific discourse is assumed to be objective. Frățilă (2007) criticises this
statement by signalling two different functions of hedging that challenge this objectivity.
First, she says that "[the authors] use hedges to distance themselves from their statements
and thus to reduce the risk of opposition and minimize the threat to face that may be part
of any act of communication" (Frățilă 2007: 89). Frățilă (2007: 89) sees hedges as threat-to-
face minimisers, which are useful techniques to express mitigation of responsibility and
uncertainty towards a proposition "in order to place their users in a secure corner, in case
other people have different opinions of the same issue and to somehow negotiate some
degree of flexibility for their claims." The distancing from a statement is reminiscent of the
negation of responsibility, one of the functional motivations for evidentiality described
above. In addition to the use of hedges to forestall criticism, hedges are also treated as a
politeness strategy. They help to build a polite relationship between the author and the
reader since hedges soften strong statements and/or straightforward opinions and at the
same time reduce possible disagreement. Moreover, since adopting a polite attitude
towards fellow scientists is essential, hedges are useful to maintain a good relationship
between scientists in the scientific community. This politeness can be expressed by suitable
quotatives when referring to colleagues' opinions. Thus, according to Frățilă, the two
functions of hedges express subjectivity, which undoes the myth of objectivity in a
scientific discourse. Additionally, it is impossible to detach the author and his/her personal
opinion from his/her writing. (Frățilă 2007: 89-91)
In sum, if one evaluates hedges as some kind of evidentials, one may assume that the
functional motivations for evidentiality expressed in the previous section are also
applicable to, and even more meaningful in scientific writing.
28
2 Audiovisual translation
The aim of this dissertation is twofold. First of all I investigate the notion of evidentiality (as
discussed in section 1) in a corpus consisting of English data from the documentary How the
Earth Was Made. Secondly, the Dutch subtitles of those English evidential statements are
studied. In order to come to the investigation of the evidential translation from English into
Dutch, one has to overcome two difficulties associated with the chosen medium.
In contrast with written texts, the present corpus offers an extra dimension. There is
no one-to-one relationship, like for instance with a written text which is translated into a
target language and again published in a written form. A documentary starts with a script
which is interpreted (mainly) by a narrator. This first transformation, from the script into
the narrator's voice, is what the viewers hear during the documentary. Secondly, this aural
information is then, after being translated, transferred into visual information via subtitles.
Thus, with this multimodal medium a lot more aspects have to be taken into account, which
will be outlined in section 2.1.1.
Because of this multimodality, the second part (section 2.1.2) of this main section is
not a mere study of how statements can be translated, but rather of how this audio-visual
transfer (into subtitles) influences the translation of the evidential expressions. Because of
the value of evidentials in scientific texts, one would expect a faithful translation of these
markers. Surprisingly, this is not always the case. We will see that it is important to
differentiate an infelicitous translation due to the different expressions of evidentiality in
the source and target languages from an infelicitous translation due to the techniques of
subtitling. The former is illustrated in example (1), taken from Taylor's (2004) paper about
subtitling. He uses the example to explain the term decimation and at the same time shows
(unconsciously) the effect subtitling can have on evidential expressions. Decimation occurs
when part of the semantic content of the original expression gets lost.
(1) You will have heard on the news that all the passengers were killed.
Lo sai che tutti passegeri sono morti.
'You know all the passengers are dead.'
(Taylor 2004: 159)
The original English sentence which expresses a quotative evidential (they said it on the news)
is subtitled into Italian. In Italian the quotative has disappeared and by using Lo sai/you
know, the indication of how the information is received has become insignificant. The
29
example above shows how the translation strategies used in subtitles, in this example
decimation, can have an influence on the expression of evidentiality and consequently on
the investigation of the corpus. Section 2.1.2 elucidates more strategies of how subtitles can
affect the (evidential) structure and/or message of the source language.
2.1 Subtitling
2.1.1 Definition
All studies (e.g. Díaz Cintas 2008; Gambier & Doorslaer 2010; Gottlieb 2004; Ivarsson &
Carroll 1998; de Linde & Kay 1999; Orero 2004; Taylor 2004) agree that subtitling is a
relatively new practice which suffered an increase in the late 20th century when
audiovisual media boomed. It is defined by Gottlieb as
the rendering in a different language of verbal messages in filmic media,
in the shape of one or more lines of written text, presented on the screen
in synch with the original verbal message.
(Gottlieb 2004: 86)
This definition is only relevant for interlingual subtitling, i.e. subtitling between different
languages, "as opposed to the intralingual type aimed at deaf and hard-of-hearing
audiences" (Gottlieb 2004: 86). Accordingly, the intralingual type is not taken into account
in this investigation.
In addition, there is an agreement that subtitling has to be treated differently from
normal standard translation, due to the multimodal nature of subtitling. Multimodality is
introduced by Jakobson as intersemiotic translation and is characterised by
the fact that the meaning created in one modality (e.g. the visual) may
be translated in another modality (e.g. the written language) or even
vice versa in this digitally manipulative age, but it also simply means
that the source and target texts consist of a number of interacting
semiotic modalities.
(Jakobson 1966 qtd. in Taylor 2004: 157)
30
One can see that the intersemiotic nature of subtitling, from speech to writing, contrasts
with the isosemiotic nature of 'normal' translation, from writing/speech to writing/speech
(Gottlieb 2004: 86).
Next to the intersemiotic difference, subtitling has another distinctive feature. With
the translation of a polysemiotic text one has to be aware of the non-verbal channels: the
visual and audio components like music and effects, picture and writing (displays and
captions) (Gottlieb 2004; de Linde & Kay 1999; Taylor 2004). If these additional components
carry (part of) the meaning, omissions in the target language are possible. Consequently,
the semiotic modalities other than the verbal can facilitate the processes of deletion.
2.1.2 Strategies
It is now clear that subtitles can be treated as a particular form of translation. The special
strategies used by subtitlers are discussed in this section.
According to Jan Ivarsson and Mary Carroll (1998: 85-91), a subtitler uses some
specific techniques being (1) condensing the text, (2) omission or paraphrase, (3) muddled
speech, (4) ellipsis, (5) merging short dialogues, (6) simplifying the syntax , (7) simple
vocabulary, (8) subtitle breaks, (9) borrowing time. The explanation of the techniques starts
with (3) muddled speech and (5) merging short dialogues. When dealing with dialogues, it
is advisable to merge two or three phrases expressed by the same person into one single
subtitle. Muddled speech occurs most of the times with interviews. It may happen that an
interviewee speaks very unclear, with incoherent phrases and bad articulation. In such case
"the best approach is to (...) translate relatively freely, giving the gist of what the person has
said in a reasonably coherent form" (Ivarsson & Carroll 1998: 87).
Nevertheless, even though the speakers explain things as accurately as possible, this
does not mean they use simple vocabulary. It is obvious that "it is easier for viewers to
absorb and it takes them less time to read simple, familiar words than unusual ones"
(Ivarsson & Carroll 1998: 89). That is why, generally, a subtitler opts for the more common
word when dealing with a choice of two synonyms. But this is only done when it does not
affect the style and spirit of the genre. In the corpus analysis it will become clear whether
the informative and scientific style of the documentary influences the vocabulary of the
subtitles. (Ivarsson & Carroll 1998)
The other techniques can be subsumed under deletion strategies or time strategies.
To begin with, according to Ivarsson and Carroll (1998) condensing the text is the most
31
difficult task of a subtitler. When the characters in, for instance a movie, talk a lot more
than there is space for to translate in the subtitles below the picture, the text has to be
condensed, which means that the subtitler has to decide what to translate and what to
delete. This decision should be based on what he thinks is the essential content for the
viewer to understand the message. This might be the first reason why evidential markers
disappear in subtitles, as illustrated in Taylor’s example in 2.1.1.
After one has decided which content should be transmitted, one has to choose which
technique to adopt: (2) omission (cf. example 2) or paraphrase (cf. example 3).
(2) "Well, it's just that this morning Mr. Smith came into the office and told us
that he has heard that we will all be fired some time during next week."
This morning Mr. Smith told us that we will all be fired next week.
(Ivarsson & Carroll 1998: 86)
(3) "You should have heard what Doris said. It appears that 'She goes out with
American soldiers' said behind your back means you're well on the way to Hell
and Damnation!"
'Going out with an American soldier', said Doris, 'is a shortcut to hell.'
(Ivarsson & Carroll 1998: 86)
According to Ivarsson and Carroll (1998), there is no strict rule to adopt the one or the
other. On the one hand they prefer omitting part of the text, because they believe
"[o]mission is less intrusive than paraphrase, a point worth bearing in mind, especially if
the original is an artistic work or a statement by a person responsible for the opinions
expressed" (Ivarsson & Carroll 1998: 86). Taylor (2004) follows the idea that deletion is the
most commonly used technique by subtitlers, but he attributes this to the intersemiotic
nature of subtitles. On the other hand, Ivarsson and Carroll (1998: 87) admit that "[m]ost
subtitlers do not consciously choose between these approaches: omitting part of the
dialogue usually makes it necessary to paraphrase the rest to a certain extent."
Another important factor, analogous with the previous one, is ellipsis. Ivarsson and
Carroll literally say the following:
Words whose main purpose is to keep the conversation ticking over
("well", "you know"), tautologies and repetitions can safely be omitted.
But this does not mean to say that subtitlers should ignore those little
32
words that often make all the difference or give the lie to a person's
character.
(Ivarsson & Carroll 1998: 87)
They warn for the danger of ellipsis. Bearing in mind the functions of evidentials, one could
almost interpret 'words that often make all the difference or give the lie to a person's
character', as a feature of evidentials. Like in the constructed example of (4), the ellipsis of
evidential markers has to be taken into account, because otherwise the subtitle is a lie
uttered by the speaker.
(4) The ancient Greeks believed that the earth was flat.
De aarde was plat.
'The earth was flat.'
Apart from the techniques that deal with omission, one also has to consider the syntax and
vocabulary. The sixth technique, simplifying the syntax, can also have an influence on
whether or not one expresses evidentiality. It is clear that "[s]imple syntactic structures
tend to be shorter than complex ones, and the difference in terms of meaning is sometimes
negligible" (Ivarsson & Carroll 1998: 88). By re-evaluating example (4), it seems obvious that
the sentence with an embedded clause is longer, due to the complex syntactic structure,
than the simple sentence used in the subtitle. Even though the latter might fit more into the
limited space of the subtitle area, it does not give an accurate representation of the
connotation (it's their opinion) meant by the speaker.
The penultimate technique mentioned by Ivarsson and Carroll is subtitle breaks.
This is an important notion for the subtitler him/herself. In short it means that the subtitler
has to avoid making subtitle breaks between structures that belong together. The goal
should be to "make the breaks coincide with the beginnings and ends of phrases, so that
each subtitle forms a coherent unit"(Ivarsson & Carroll 1998: 90), as seen in example (5).
(5)
(Ivarsson & Carroll 1998: 91)
Wrong Right
She had furnished the room well.
The interior was mainly in red
She had furnished the room well
and green, as these were her
favourite colours.
The interior was mainly in red
and green, her favourite colours.
33
In the explanation of the last technique, borrowing time, Ivarsson and Carroll admit again
that evidential markers can get lost in subtitling. The instances they denote as small talk
may express (un)certainty, which is central in English evidentiality:
Where absolutely necessary, subtitlers can "borrow time" from the next
sentence (...). This means letting the subtitle remain on the screen
during the first part of the next sentence. The dialogue that is masked in
this way should of course be of minor importance. Fortunately,
sentences often start with small talk such as "Well, I don't know really, I
suppose you could say ...". Such phrases can be omitted or abridged.
(Ivarsson & Carroll 1998: 91)
Apart from all the previously mentioned techniques of subtitling, another notion to bear in
mind is punctuation. Especially in the case of evidentiality, the right use of quotation
marks is essential. Also in subtitles, they cannot be omitted, "[t]hey should be repeated in
each of the subtitles, otherwise viewers tend to forget that they are still reading a
quotation" (Ivarsson & Carroll 1998: 114). Quotation marks are necessary when quoting a
particular source, being in most cases a person. However, in a documentary (or television in
general) it is possible to view the person who is speaking. Consequently, literal evidentiality
markers, like quotation marks, disappear because of the combination of the verbal with the
visual medium, like I have already pointed out in the beginning. (Ivarsson & Carroll 1998)
34
II: CORPUS ANALYSIS
3 Methodology
3.1 Data collection
The corpus used for this investigation is made up of sentences found in the American
television documentary How the Earth was Made. The data are taken from the first season,
which comprises thirteen forty-five minute episodes. This popularised genre of science was
chosen because of the following reasons. First, evidentiality is centered around the
expression of the source of information. Acknowledging the source to explain certain
findings is an important feature of science as well. Second, the television documentary
offers an additional advantage, since the Dutch subtitles can help to define the English
evidentiality. The corpus was created as follows. Each episode was watched attentively and
every evidential-like expression was written down together with its concomitant Dutch
subtitles. American English evidentiality thus served as starting point. Afterwards, the
utterances were re-examined and only the true evidential statements were retained.2 As a
result, the corpus consists of 512 tokens, 255 of which are English sentences containing an
evidential marker, spoken by the narrator or persons in the documentary.
The episodes of the documentary follow a fixed format, which emphasises the
scientific nature of this popularised form of science. According to Meyers (2003: 273), "[t]he
dominant model of popularization assumes that the aim of the process is to convey
scientific knowledge to a wider audience." The composition of each episode contributes to
the achievement of this goal. For instance, the second part of the fourth DVD deals with the
origin of Hawaii. As an introduction, the narrator gives a description and the geological
facts of Hawaii, containing the biggest volcano on earth. Next, the first question, “Could this
huge island have been built entirely by lava?”, is asked, followed by clues, research and
opinions of scientists, which lead to an answer. When the mystery is solved, the question is
answered and followed by the enumeration and appearance of the most important clues on
the screen. Subsequently, the next question, “What force on earth could create the most
active volcano in the world?”, is uttered and the process of investigation is recapitulated. At
the end, the key information is repeated in order to offer a complete account of how the
2 The criteria used to compose the corpus are outlined in section 3.2.
35
Hawaiian chain was created. In this way, all the episodes are clues to answer the main
question of how the earth was made. This approach of repetition, summaries, question-
answer, and experts who search for and provide evidence, is reminiscent of scientific
investigation and serves the aim of popularised science. Consequently, in this scientific
context the indication of the source of information and the expression of (un)certainty
plays a crucial role.
3.2 Criteria
The selection of the data was based on the functional definition of English evidentiality,
outlined in the literature review. The point of departure for the recognition of the
evidential forms were the formal devices expressed in Figure 2 (cf. section 1.2.2), which
stand for the two meanings, i.e. indication of information source and expression of
(un)certainty. While watching the documentary, the question was asked whether certain
utterances indicated direct evidence, reported evidence or inferential evidence, or whether
they expressed certainty or uncertainty.
The recognition of sentences that signal direct and reported evidence turned out to
be uncomplicated, whereas the inferential evidentiality type caused some difficulties. For
this particular study it is advisable to reconsider English inferential expressions. As
explained previously, while watching the documentary it becomes clear that all the
episodes are centred around evidence. Evidence is given in order to answer the main
question asked at the beginning, which deals with the origin of a certain natural
phenomenon. This evidence is presented by means of evidential markers, which are
especially based on inference. In the literature review, de Haan (2001: 193) defines inference
as "the grammaticalized way of showing that the speaker makes his/her statement based
on a deduction from facts, and not on a direct observation of the action itself." Examples
(1a) and (1b) are the sentences that accompanied this definition. In (1a), which is an
instance of Tuyuca, the speaker inferred the statement from the available evidence whereas
in (1b) the evidence on which the inference is based is expressed overtly. Example (1b) is an
instance of Tepehuan, a Uto-Aztecan language which uses particles to express inference.
(1a) bóahõã-yu
rot-INFER:OTHER:PAST
'It rotted' (said of a plant after pulling it up to examine it)
(de Haan 2001: 193)
36
(1b) Dáman dara-t vac dyi turasno.
shallow sit-PI INFER ART peach
Pui' cu-t va' ia'ray gu jivihl.
Thus so-PI then. fell ART wind
'These peach trees must have been planted shallowly. That's why the wind
blew them over'.
(de Haan 2001: 193)
These examples demonstrate two kinds of inference. In (1a) the evidence is unspecified,
whereas in (1b) the speaker refers to a specific inferential source. In addition, de Haan
focuses on inference as a grammatical phenomenon. Thus, in the American Indian
sentences above the fact that the speaker has evidence is expressed either by a verbal suffix
(1a) or by a particle (1b). In English, by contrast, it is necessary to express evidentiality
lexically in order to obtain an accurate representation of the connotation meant in the
grammaticalised evidentiality language. Therefore, it is impossible to recognise the kind of
evidentiality expressed by example (1a) in English. The second kind of inference, by
contrast, is used abundantly in the documentary. In example (1b) above the available
evidence (the fact that the wind blew them over) is expressed. In other words, example (1b) can
be interpreted as: The wind blew the peach trees over. This is evidence that the peach trees must
have been planted shallowly. In this view, I would like to examine the following four tokens:
(2) Gray gneiss rocks at the top of the Matterhorn prove that Africa collided with
Europe, creating the Alps.
(3) The crystals are clear evidence that under the springs is a volcano.
(4) Diatomite reveals the climate was once wetter.
(5) What's striking about these samples is that the minerals inside are elongated.
It is a clue that these rocks must once have been crushed by massive forces.
In example (2) the statement is: Africa collided with Europe, creating the Alps. The speaker has
inferred this statement from the available evidence and the evidence is expressed overtly,
namely the gray gneiss rocks at the top of the Matterhorn. Therefore, Gray gneiss rocks at the top of
the Matterhorn are evidence that Africa collided with Europe, creating the Alps is a fitting
paraphrase. Similarly, following the interpretation of (1b), example (3) can be paraphrased
as I see crystals under the springs. I have evidence that under the springs is a volcano. The same
goes for example (4): The presence of diatomite is evidence that the climate was once wetter.
37
Finally, sentence (5) coincides almost completely with the translation of (1b) because of the
use of the evidential must. Consequently, these sentences are included in the corpus as true
evidential expressions, in particular as evidentials that denote inference.
3.3 Methods
The corpus analysis consists of two studies. The aim of the first analysis is twofold. The
study intends to shed light on the features and the occurrence of evidentiality types in a
specific genre, namely scientific narrative. In this way, this corpus-specific investigation
offers a broader view on the concept of evidentiality in the English language on the one
hand, and serves as a basis for the comparative study that follows, on the other hand.
The first study deals with the English evidential expressions in a qualitative and in a
quantitative manner. First of all, the data were classified with regard to the type of evidence
they express, i.e. direct evidence, reported evidence, inferential evidence and additionally,
the expression of (un)certainty. Consequently, the categorisation was made with the
following questions in mind: Did the speaker witness the event? Did the speaker hear the
statement? Was the statement inferred from available facts? Does the speaker express
(un)certainty concerning his/her statement? Section 4.1.1 presents the results of the
classification by outlining the difficulties encountered when analysing the data and by
discussing specific features of each type. Once the classification was done, the frequency of
each type was calculated and presented in Table 1. Thirdly, as opposed to the semantic
approach in the first classification, section 4.1.2 focuses on the evidential expressions from
a formal point of view. In this section the evidentials are divided according to their internal
structure, with an emphasis on the verb phrase (VP). Finally, to conclude the first analysis
the question is raised what are the functional motivations for the evidentiality data (section
4.1.3). When analysing the genre-specific data, it was noticed that the functional
motivations expressed in the literature review were unusable. Therefore, the broad
discourse-oriented approach of Martin and White (2005) was adopted. Via their engagement
system, the functional motivations, which are led by the communicative needs of the genre,
are discovered and exemplified.
When the delineation of the English scientific evidentiality system and the
examination of the peculiarities were done, the second analysis took place. The second
study looks at the Dutch translations of the evidentiality expressions in order to gain more
insight into the English concept of evidentiality. This goal is achieved by an interlingual
38
comparative analysis of the original spoken English sentences and their concomitant
subtitles (section 4.2). To begin with, the subtitles which translate the English evidentiality
expressions literally and the ones which are marked by a difference were counted. From the
137 literal translations, the typical Dutch evidentiality translations were deduced. The other
118 tokens were further analysed and two main classes were discovered. In the first place, it
seemed that some Dutch subtitles express a change in type. Secondly, it was noted that the
subtitles are characterised by specific formulations. Explanations for these different
formulations were found and outlined in section 4.2.2 and 4.2.3. The former consists of a
detailed subtitle analysis in order to examine the role of the subtitling techniques on the
drastic differences in formulation. The latter investigates whether or not the changes are
functionally motivated. These last sections show that the comparison between the Dutch
subtitles and the English expressions does not only help to describe English evidentiality in
scientific discourse, but also to give an insight into the work of the subtitler.
39
4 Results
4.1 Classification and investigation of the English data
The current section deals exclusively with the 255 English evidentiality expressions and
contains three main subsections. The general numbers of the division into the different
evidentiality types are followed by a detailed analysis of those types. This analysis uses the
various kinds of verbs as a starting point. The pure evidentiality types, i.e. those which
indicate direct evidence (section 4.1.1.1), reported evidence (section 4.1.1.2) and inferential
evidence (section 4.1.1.3), are completed with the analysis of expressions that indicate
(un)certainty (section 4.1.1.4). After the semantically-oriented part, the second part focuses
on the formal encoding (section 4.1.2), whereas the third examines the functional
motivations for evidentiality which are at work in the documentary (section 4.1.3).
4.1.1 The English evidentiality types
TABLE 1. Incidence of types of evidence
The 585 minutes of television resulted in a corpus of 255 American English evidential
expressions and their subtitles, which gives an average of 19.62 examples of evidentiality in
each episode. The current section presents the numbers of the different evidentiality types,
which are visualised in Table 1. The diagram represents all data, including the exception
0
10
20
30
40
50
60
70
80
90
Direct evidence
Reported evidence
Inferential evidence
(Un)certainty
Combinations
Exception
40
(1) A tsunami isn't over in just a few seconds.
Men denkt dat een tsunami zo voorbij is.
Firstly, the indication of direct evidence, the distinction between visual and non-visual
evidence not taken into account, represents 52 occurrences (20,39%) on a total of 255. The
two main categories are the expressions that indicate reported and inferential evidence,
which contain 78 (30,59 %) and 81 (31,76%) examples respectively. Both types can be divided
into two subcategories. Following Aikhenvald (2004), the reporting evidentials distinguish
between quotative and hearsay evidence, while the inferentials indicate inference and
assumption. Quotative evidentials refer overtly to the quoted source whereas hearsay
evidentials do not. The reported evidence type is almost equally divided into quotative
(37/78) and hearsay (41/78) evidentials. By contrast, within the inferential evidence type
one finds four assumptions in comparison with 77 inference examples. Secondly, the
expressions of (un)certainty count 38 tokens. Finally, five sentences represent more than
one type.
4.1.1.1 Direct evidence
The first type to be described is the one that indicates direct evidence, which means the
speaker was a direct witness of the event. This direct observation can be visual, non-visual
or based on memory. The corpus comprises 51 tokens of direct evidence, all signalling
visual evidence. Normally, the visual direct evidence is expressed by perception verbs, but
in the corpus the most frequently used direct evidence verbs are to show and to reveal. The
former is a special type of perception verb, namely a causative perception verb: 'To show
something to someone' means 'to make someone see something'. Similarly, the latter has a
causative connotation when used with a clause, being 'cause or allow (something) to be
seen'.3 Example (2) illustrates the visual meaning literally by using the subject footage. In the
translations also, a link between to show/to reveal and to see is established by translating
them as laten zien or onthullen, which makes the causative meaning explicit.
(2) This incredible time-lapse footage shows that because the top of the flow is
still exposed to the air, it cools and crusts over.
3 Definition is quoted from the Oxford Dictionary, consulted online: http://oxforddictionaries.com/definition/english/reveal?q=to+reveal 25/04/2013
41
Only five expressions that indicate direct visual evidence are characterised by the verbs to
see and to notice. In the corpus, to show is exclusively used with an inanimate subject, while to
notice and to see require a human subject. This distinction is further developed in section
4.1.2.
4.1.1.2 Reported evidence
78 out of 255 evidentiality expressions mark reported evidence, including both quotative
and hearsay evidence. The present section introduces the findings noticed when analysing
this second largest group of data. Specifically, it discusses the difficulties observed when
categorising the reporting evidentials, as well as the specific features of these evidentials in
the genre of television documentaries.
First of all, while dividing the data into the four categories, a case of ambiguity
occurred. Data (3), (4), (5) and (6) serve as illustration:
(3) Samurai writings speak of people living on the coast running for higher
ground as soon as they felt an earthquake.
(4) Native American myths explain the earthquake as shocks from a battle
between warrior spirits.
(5) Research suggests we might not be as safe as once thought.
(6) This rock is telling us that we had basically wet land.
The examples above challenge the notion of reported evidence on the one hand, and
inference on the other hand. When evidentials signal reported evidence they usually
indicate a human source, whereas inferential evidentials refer to a deduction from facts. As
seen in the literature review (cf. Figure 2), the former is expressed in English by people say,
they suggest, he is reputed, etc., in other words, by verbs of communication. When looking at
sentences (3) to (6), the condition of 'saying verbs' is fulfilled, which, at first sight, justifies a
classification as evidentials that indicate reported evidence. However, in none of the
examples does a person occur who reported the statement; all the subjects designate
inanimate things. Consequently, the question is raised whether these statements may be
based on inference. If the answer to this question is positive, the Samurai writings, the Native
American myths, the research and the rock, were investigated by the speaker and he/she
inferred the statements from them. The two options, i.e. reported evidentials or inferential
evidentials, are plausible but unsatisfactory, since a closer look reveals significant
42
distinctions between the subjects. From a poetic point of view, the lifeless Samurai writings
that speak of something are an instance of personification. In this view, data (3) to (6) are all
examples of personification. However, the attribution of human qualities to something
nonhuman is more acceptable with examples (3) and (4). In sentence (3), for instance, the
subject Samurai writings implies a human activity, i.e. someone writing something.
Accordingly, the subject does express a person, albeit implicitly. As a result, this example of
personification can be rephrased as: The Samurai speak (in their writings) of people living on the
coast running for higher ground as soon as they felt an earthquake. The subtitler agrees with this
interpretation and transfers the nonhuman subject into the Samurai and the saying verb
into a verb of description: De Samurai beschrijven dat kustbewoners naar hogere gebieden
vluchtten, zodra ze een aarschok voelden ('The Samurai describe that people living on the coast
running for higher ground as soon as they felt an earthquake'). The Native American myths of
illustration (4) require a similar interpretation because the source of myths is typically
human, and because this subject is reminiscent of Anderson’s subdivisions (cf. section 1.1.2).
He included myth and history into the type of reported evidence.
Because of the implication of a human (activity), this dissertation prefers to treat
examples (3) and (4) as belonging to the type of reported evidence. Examples (5) and (6)
need a different analysis. The former can be interpreted as partly reporting, partly
inferential whereas the latter only accepts an inferential reading (From the features of this
rock we deduce we had basically wet land) due to the totally nonhuman nature of the subject
this rock. The following sentences illustrate the double interpretation of sentence (5):
(5) Research suggests we might not be as safe as once thought.
'Researchers suggest we might not be as safe as once thought.'
'From the facts represented in the research I deduce we might not be as safe
as once thought.'
Both analyses are valid because a research implies the investigators who did the research,
as can be seen literally in example (7), and at the same time the research represents facts
deduced from a scientific investigation.
(7) New research by geophysicist Ares Rosakis suggests that the San
Andreas may offer a new and even more deadlier threat.
43
Due to the scientific features of a research, the inferential interpretation is more sufficient
than with the writing of myths. As a consequence, evidentials (5) and (6) are considered in
this dissertation as inferentials. To sum up, similar instances of personification are
evaluated according to the level of humanity or investigation implied in the source.
After having delineated what constitutes the reported evidence type, the remaining
part of this section focuses on the subdivisions of the reporting evidential, i.e. hearsay and
quotative evidence. Of the 78 data of reported evidence, 37 mark quotative evidence while
41 denote hearsay evidence. Given that the aim of popularised scientific discourse is to
involve people from outside the scientific community, this almost equal division between
referring to the specific source and not specifying the source, is not remarkable. The
laypeople who are watching the documentary, are interested in the results of the research
itself, not in the names of the scholars behind it of whom they have never heard. This
explains the high number of hearsay uses. Another possible explanation for these almost
equal numbers may be the frequent use of generic nouns. The narrator in the documentary
gives a lot of information via the use of the plural, which refers to a whole group or class.
Example (8) shows this generic approach and contrasts with example (9):
(8) Many geologists believe the exact same volcanic action accounts for the
formation of the fifth and final lake.
(9) Geologist Mike Poland believes they're major clues in the mystery of Iceland's
formation.
Since generic nouns refer to a whole class (of geologists), the subject of sentence (8) is
considered vague enough to be analysed as hearsay evidence. In example (9), by contrast,
the source cannot be described more specifically. This specific description is due to the fact
that, in most of these quotative cases, the person mentioned by the narrator actually
features in the documentary. In this view, example (9) serves as an introduction of the
geologist Mike Poland.
Finally, seven of the 78 reporting evidentials are realised by alternative means, like
for instance, quotation marks. With regard to quotation marks both quotative (cf. example
10) and hearsay evidence (cf. example 11) are represented.
(10) The pilot said: "It seems that we're in some sort of a chimney."
(11) The question then became: "What drives the hotspot? What causes the
hotspot?"
44
The small number of quotation marks reveals a tendency towards indirect speech in science
documentaries and is a consequence of the multimodal medium. Next to the introduction of
an expert by a quotative sentence uttered by the narrator, the makers of the documentary
also acknowledge the source by showing the name and profession of the speaker on the
screen. Because the speaker is shown and the name is presented, quotation marks are not
needed. These examples are not included in the corpus.
4.1.1.3 Inferential evidence
In view of the criteria proposed in section 3.2, most inferential evidentials4 are
characterised by verbal expressions such as to prove, to reveal, to turn out, to indicate, to
conclude and this is proof, this is evidence, these are clues, these are indications. Yet, none of these
devices was mentioned by Ifantidou (cf. Figure 3), who indicates I gather, I guess, I suppose, I
deduce as typical inferential verbs. The lack of these inferential verbs in the corpus is due to
the distinct genre. As can be deduced from the use of the first person singular, Ifantidou's
verbs were derived from spontaneous conversations. The verbs, which emphasise the
speaker, show that the direct speaker-hearer interaction is central. In the corpus, by
contrast, sentences like the constructed example (12b) are not found.
(12a) Those high pressures indicate that these rocks formerly were produced at
depths of 20 to 25 miles.
(12b) 'I deduce from those high pressures that these rocks formerly were produced
at depths of 20 to 25 miles.' (constructed example)
The verb to deduce, together with to guess, to gather and to suppose, only allows a person as
subject of the sentence. But, in the corpus, human subjects followed by an inferential verb
are non-existent. The subjects of this inference group refer to things, certain tests,
experiments or investigations on which the statements are based. The preference for these
inanimate things as subjects, which is the emphasised part of the sentence, has two
explanations. First of all, the scientific nature of the expressions plays an important role.
With a documentary which is centred around seeking and providing evidence, it is obvious
that the evidence needs most attention, and this is achieved by putting it at the beginning
of the sentence. Secondly, most of the inferential data are spoken by the narrator, who does
not act in the 'story' he is telling. It is his job to stay in the background, so he only gives an
4 inferential here means the evidentials based on inference, not on assumption.
45
impersonal objective account of the facts. Thus, the conversational interaction which marks
Ifantidou's verbs does not take place between viewer and documentary maker nor between
viewer and narrator.
Nevertheless, the narrator does introduce names of scientists, but only as the
authors of their research or findings. Examples are the following:
(13) Bauder's measurements reveal this glacier moved over ten feet every month.
(14) Morgan's research proves that geysers were exploding around the lake.
(15) Doctor Roberts' data reveal that this whole glacier is sliding forward at an
astounding rate of two feet per day.
In sum, Ifantidou's verbs require a person who acts in a conversation, while the inferential
data show a tendency towards impersonalisation due to the evidence-based and objective
nature of the science documentary.
In contrast with the inference group, the second subdivision of the inferential type
does accept human subjects. The information above discusses inferential evidentials which
indicate inference, while the part that follows deals with assumption. Only four cases of
assumption are comprised in the corpus:
(16) Scientists assumed the destruction must have been caused by an asteroid of
at least 100 feet.
(17) Throughout most of recorded history man has just assumed that beyond
certain level the sea was pretty flat, and fairly lifeless.
(18) Gilbert assumed that if the crater was caused by an asteroid, he should find a
giant alien rock in the middle of it.
(19) The assumption was there had to be another volcano these days.
In examples (16), (17) and (18) above, the person who assumes is expressed. This is obvious
given that the verb to assume follows the pattern 'someone assumes something'. Alternative
patterns exist. To assume can be replaced by the assumption is which again expresses the
typical scientific impersonalisation through nominalisation, as noted in example (19).
Both inference and assumption may be expressed by the modal must. The inferential
meaning of must have/must be is illustrated clearly in the following data:
(20) The grooves in the rocks could mean just one thing. The glacier must have
been thousands of feet thick
46
(21) Yellowstone's hot water features all point to one conclusion.
Yellowstone must be powered by the heath of a volcano.
These data are reminiscent of example (1b) of de Haan, mentioned in section 3.2. First the
evidence is given and in the second sentence the findings are inferred and expressed by use
of the modal must. This observation is in favour of the proposition made in the literature
review: There is smoke in the kitchen. The toast must have burned. Consequently, there is no
doubt that modal must may be treated as an evidential. In addition, all the examples
expressed above have a high level of certainty because of the use of must. More about the
second meaning of evidentials can be found in section 4.1.1.4
Although the verbs proposed by Ifantidou are not present in the corpus, the devices
treated by Ifantidou as borderline cases, are. For instance, the conjunction so expresses an
evidential meaning in examples (22) and (23). Its meaning is often reinforced by the
evidential must.
(22) They are made of coral which can only grow under water. So some immense
force, unleashed by Krakatoa, must have put them here.
(23) Today, you see here hills and trees. So, obviously, there's not a big hole left in
the ground from the eruption of the Yellowstone volcano.
Finally, a small number of the inferential verbs is represented by the verbs to tell and to
confirm. These verbs denote cases of personification with subjects such as these bubbles, ice
cores from Antarctica, thick layers of debris, etc. As already discussed when dealing with
reported evidence, they are analysed as inferential evidentials because of their nonhuman
or scientific features.
4.1.1.4 Expression of (un)certainty
Next to verbs of seeing, communication or inference, the corpus consists of a vast amount
of verbs that express (un)certainty such as to suspect, to know, to doubt, to think, to have a
hunch. After having examined the basic evidentiality meaning, i.e. indicating information
source, which is outlined above via the three types, this section turns to the second
meaning, the expression of certainty or uncertainty. Particularly, these expressions
represent 38 occurrences of the complete 255 tokens and are characterised by a 'verb of
doubt', a 'verb of certainty', an epistemic modal, or adverbials like obviously.
47
As explained in section 3.1, the television documentary consists of a combination of
scientists talking about their research (e.g. I believe), and a narrator who summarises their
findings (e.g. scientists think). Usually the narrator, via a voice-over, starts with the doubts or
beliefs of one or more scientists, which he expresses with sentences like (24). Afterwards,
their ideas are substantiated with evidence and the uncertainty verbs change into verbs
expressing certainty, often accompanied by the adverb of time now:
(24) Scientists had a hunch that this colossal ridge might help explain how the
trench was born.
(25) We now know, by looking at meteorites like this one for example, that in fact
the solar system was formed 4,567 billion years ago.
Most of the (un)certainty types establish a link with one of the information source types. By
analysing example (26) below, one notices a combination of inference (so) and certainty
(obviously) while sentence (25) is a clear example of how the certainty expression (we now
know) is based on observation (by looking at meteorites). Examples (27) and (28) show how
(un)certainty expressions may also indicate hearsay evidence, mostly expressed by a
passive voice. What is more, as already mentioned, modal must is used both for the
indication of certainty as well as for the indication of inference, as is illustrated by example
(29).
(26) Today, you see here hills and trees. So, obviously, there's not a big hole left in
the ground from the eruption of the Yellowstone volcano.
(27) Originally these were thought to be related to volcano activity.
(28) It is now known what a geological wonder the Mariana trench is.
(29) Based on the size of the boulders that we see here, we know that that float
must have discharged down the order of 1.5 million cubic feet per second.
These (un)certainty types which also express the source of information reinforce the
statement made in the literature review, which said that in non-grammatical evidentiality
languages the expression of evidentiality frequently coincides with epistemic expressions.
These findings justify the inclusion of the expression of (un)certainty as second meaning of
evidentiality.
In addition, the adoption of the pragmatic inference technique is possible on most
(un)certainty data. In sentence (30) the narrator says that Tankersley is convinced.
48
(30) Tankersley is convinced catastrophe drove the mammoths to extinction.
If the pragmatic inference technique is adopted, one may deduce that the speaker knows
Tankersley is convinced because Tankersley himself told it to him (reported evidence), the
narrator inferred this conviction from research by Tankersley (inferential evidence) or the
narrator deduced it from his convincing behaviour (direct evidence). Sentences (24) and
(25) allow this pragmatic inference technique as well.
4.1.2 Formal encoding
As indicated before, English expresses evidentiality lexically, by using devices such as verbs
(I hear, he says, I deduce, he knows), modal verbs (must, may), adverbs (allegedly, consequently,
probably), adjectives (evident, possible), nouns (assumption, evidence, proof), conjunctions (so)
and lexical phrases (according to). The speaker thus has a variety of choices, but this variety
is not represented in the corpus. Full verbs account for the highest incidence (213/255), and
are commonly accompanied by the following options:
1. Noun phrase (NP) + verb
The frequent earthquakes on Anak Krakatau indicate the volcano is still
dangerous.
2. Proper noun (PN) + verb
Menzies believes the core sediments the river left behind, dramatically
accelerated the ice sheet's flow.
3. Pronoun (Pron) + verb
They believe the shape of the descending tectonic plate may hold the
answer.
The first category (NP + Verb) represents most occurrences (140/213). The possibility
pronoun + verb (Pron + Verb) occurs in 43 cases, followed by the proper nouns + verb (PN +
Verb) with 30 examples. All three combinations are used in the four evidentiality types, as
visualised in Table 2 below. Note that the most frequent combination (NP + Verb) comprises
both inanimate things (theory, studies) and persons (scientists, geologists). The former may
express inference or direct evidence. As explained in section 4.1.1.3, almost all inference
statements are based on evidence, facts, studies or research, and direct evidence is based on
what is shown or revealed by a certain study. The nouns denoting humans of this specific
majority combination may refer to reported evidence or the expression of (un)certainty.
49
The proper nouns, obviously, are almost always linked with quotative evidence, as a
consequence of the definition outlined above. A smaller amount expresses (un)certainty,
while direct or hearsay evidence is absent. Two exceptional cases of inference are found
with the Proper noun + verb structure. The first one is one of the examples of assumption,
repeated here as example (31). The second one, sentence (32), contains the proper name of
a submarine, analysed according to its nonhuman features as an inferential example.
(31) Gilbert assumed that if the crater was caused by an asteroid, he should find a
giant alien rock in the middle of it.
(32) The Trieste dived to the bode of the trench and confirmed that it is the
deepest point on the planet.
Thirdly, the pronouns may refer to NPs as well as to proper nouns mentioned in the
previous sentences, which explains the occurrence of this structure in all types.
TABLE 2. Frequency of the structures NP + Verb, PN + Verb and Pron + verb
The three frequent structures as evidential markers are followed by the statement.
This combination has as a consequence that the majority (83,53%) of the data are
constituted by a subject, a main verb and a that-clause as object. Even though the
conjunction that is not always expressed, all the example presented above follow this
pattern.
Among the less frequent formal structures, one finds nouns (15 occurrences), modal
must (4 occurrences), conjunctions (3 occurrences), adjectives (4 occurrences), passive
constructions (8 occurrences), and quotation marks (7 occurrences). Quotation marks are a
non-linguistic way to express evidentiality. Obviously, when dealing with spoken language,
one cannot see the quotation marks. Nevertheless, the seven occurrences were detected
because of the typical direct speech structure of the spoken sentence, the intonation of the
speaker and the quotation marks used in the subtitles. The quotation marks in example (34)
(Un)certainty Direct Reported Inference
Quotative Hearsay
NP + Verb 16 40 10 29 43
PN + Verb 7 / 17 / 4
Pron + Verb 11 10 10 4 6
50
indicate quotative reported evidence and are accompanied by the linguistic evidential
markers NP + verb.
(33) The pilot said: "It seems that we're in some sort of a chimney."
4.1.3 Functional motivations for evidentiality
As mentioned in the literature review, the indication of the source of evidence serves the
speaker’s communicative purposes. Below, the three proposed functional motivations from
the perspective of the speaker are summed up, accompanied by prototypical data:
1. Acknowledge responsibility for a statement
We see that the secret of the slipping San Andreas Fault is actually the
rock itself.
2. Deny responsibility for a statement
Experts warn if this warming continues, the Alps will by ice-free by the
end of the century and they fear these great mountain peaks will tumble
down even faster.
3. Assign authority to the statement
Da Vinci proposed that these fossils were formed under the ocean.
Marine fossils are evidence that these rocks were once covered by a
tropical sea.
When applying these functional motivations, which were proposed in the literature review,
to the data, a problem occurred. It was noticed that these functional motivations are not
sufficient when dealing with the data of the specific scientific corpus. For instance, they do
not take into account inferential sentences like example (34). Following the functional
motivations expressed above, the speaker of the sentence would be classified as denying
responsibility for the statement because he is no direct witness. At the same time, the
sentence itself implies a high degree of certainty and conviction expressed by the speaker,
which means he does act as responsible for the statement. Because of this flaw in the
categorisation, it is recommended to start from a broader point of view, namely the
engagement types proposed by Martin and White (2005).
51
(34) The theory showed that the earth's crust consists of separate moving plates
on which the oceans and continents sit.
The first part of the analysis (section 4.1.3.1) sheds light on the engagement categories
which use an evidential marker. The second (section 4.1.3.2) tries to find the 'engagement
motivations' of the documentary maker and/or the narrator.
4.1.3.1 Engagement positions
To fully understand the communicative purpose of the speaker and the documentary maker
to use evidential expressions, a broad interpretation of functional motivations is needed.
Martin and White (2005) deal with an engagement system which is dialogically orientated.
They adopt a broad discourse-oriented approach and focus on how the speaker 'engages'
with "prior utterances, alternative viewpoints and anticipated responses" (Martin and
White 2005: 97). Their engagement system consists of several dialogistic positions. First a
distinction is made between dialogically contractive and dialogically expansive locutions.
The former contains the categories, 'disclaim' and 'proclaim', while the latter consists of
'entertain' and 'attribution'. The diagram below presents the categorisation and the main
definitions.
52
CONTRACTION
Disclaim: The textual voice
positions itself as at odds with, or
rejecting, some contrary
position.
Deny
Counter
Proclaim: By representing the
proposition as highly
warrantable (...) the textual voice
sets itself against, suppresses or
rules out alternative positions.
Concur
Pronounce
Endorse
EXPANSION
Entertain: by explicitly
presenting the proposition as
grounded in its own contingent,
individual subjectivity, the
authorial voice represents the
propositions as but one of a
range of possible positions - it
thereby entertains or invokes
these dialogic alternatives.
Attribution: By representing
proposition as grounded in the
subjectivity of an external
voice, the textual voice
represents the proposition as but
one of a range of possible
positions - it thereby entertains
or invokes these dialogic
alternatives.
Acknowledge
Distance
(Martin and White 2005: 97-89)
The analysis starts with the explanation of the engagement resources, which can all
be exemplified by the evidential data. Therefore, they can be seen as functional motivations
to use the kind of evidentiality expressions described in the corpus. The data which
illustrate the category 'entertain' and the subcategories of the categories 'disclaim',
'proclaim', and 'attribution' are presented below.
53
The first group of examples demonstrates the dialogically contractive wordings,
which means "they close down the space for dialogic alternatives" (Martin & White 2005:
103).
(35) A tsunami isn't over in just a few seconds.
(36) This would suggest that volcanic eruptions still happen, even under the
enormous weight of ice.
(37) Today, you see here hills and trees. So, obviously, there's not a big hole left in
the ground from the eruption of the Yellowstone volcano.
(38) To me that's not something that can be explained by something that's just
near the surface.
(39) Contemporary reports show that most of Krakatau victims were not killed by
these deadly avalanches.
Example (35) is the only instance of denial found in the corpus. The denial is directed to the
addressee and anticipates the "beliefs which they (sic) speaker/writer assumes that at least
some members of his/her mass audience will be subject to" (Martin & White 2005: 119). The
following extract from an advertisement placed in magazines by the British Heart
Foundation is given as example:
We all like something to grab hold of. But sometimes you can have too
much of a good thing. And a man whose tablet diet consists of double
cheeseburgers and chips can end up looking like a tub of lard. There's
nothing wrong with meat, bread and potatoes. But how about some lean
meat, wholemeal bread and jacket potatoes?
(Martin & White 2005: 118)
The denial, There's nothing wrong with meat, bread and potatoes, includes opinions or claims by
people who believe that 'there is something wrong with meat, bread and potatoes' (Martin
& White 2005). Similarly, in example (35) the narrator responds to his audience who may
believe 'a tsunami is over in just a few seconds'. This example is included in the corpus
because its subtitle expresses evidentiality, a point which will be discussed in section 4.2.3
which deals with the Dutch subtitles. In addition, denials of this corrective type "present
the addresser as having greater expertise in some area than the addressee and as, on that
54
basis, acting to correct some misunderstanding or misconception on the addressee's part"
(Martin & White 2005: 120). This goal coincides with the specific aim of popularised science,
which is educational.
In the same category of denial, the countering proposition is found. In example (36)
the scientist/speaker counters the proposition that would have been expected. The
proposition that eruptions still happen is in counterrelationship with the enormous weight
of ice. Countering examples do not use evidential markers to express their purpose, and are
therefore excluded from the analysis. Example (37), which belongs to the inference and
certainty type, is an instance of concurrence, which "overtly announce[s] the addresser as
agreeing with, or having the same knowledge as, some projected dialogic partner" (Martin
& White 2005: 122). With the adverb of certainty obviously, the speaker makes clear that the
statement that follows is the only correct interpretation of the findings.
A reporting example where the speaker is explicitly present denotes
pronouncement. In example (38) the speaker gets involved and shares his opinion, which is
clearly by to me. According to Martin and White (2005: 128) "such insistings or emphasisings
imply the presence of resistance, some contrary pressure of doubt or challenge against
which the authorial voice asserts itself." By doing so, he reduces the possibility to express
alternative opinions. The speaker has also a particular role with endorsement.
Endorsements are "those formulations by which propositions sourced to external sources
are construed by the authorial voice as correct, valid, undeniable" (Martin & White 2005:
126). Consequently, as illustrated by the inferential example (39), the speaker takes over
responsibility for the proposition and expresses his confidence towards the cited source.
Endorsement is typically represented by the verbs to show, to prove and to demonstrate.
(Martin & White 2005)
The following group of examples belongs to the dialogically expansive ones. With
these examples "the effect is to invite or at least entertain dialogic alternatives and thereby
to lower the interpersonal cost for any who would advance such an alternative" (Martin &
White 2005: 103). The inferential example (40) belongs to the category of entertain, which
contains "those wordings by which the authorial voice indicates that its position is but one
of a number of possible positions and therefore, to greater or lesser degrees, makes dialogic
space for those possibilities" (Martin and White 2005: 104). This category is considered the
default category for modality and evidentiality. By expressing uncertainty (I think) or
surmise (apparently, to suggest, it seems) the speaker presents the propositions as one among
55
potential alternatives (Martin & white 2005). In example (40) the verb suggest indicates that
the speaker is willing to hear alternative options.
(40) New research by geophysicist Ares Rosakis suggests that the San Andreas
may offer a new and even more deadlier threat.
(41) People say that more than half a million people died from the drought in
India.
(42) The politician claims that only 2.000 people died from the drought
in India. (constructed example)
Note that I believe refers to 'entertain', while they believe or he believes is seen as an
acknowledgement. In contrast with endorsement, the author of an acknowledgement is
considered neutral. The speaker only reports, by means of reporting verbs such as to say, to
believe, to announce, without adding a personal value to the proposition. As a consequence,
the writer is presented as an "informational fair trader" who simply reports the opinions of
others without taking a stand (Martin & White 2005: 113). Nevertheless, the reader is able to
value the statement as reliable and credible or as unreliable and doubtful (Martin & White
2005). For example, the subject people in the reporting example (41) may be seen as less
reliable than for instance the subject The United Nations.
Another subcategory of the 'attribution' category is distancing, which is the
complete opposite of endorsement. The speaker 'distances' "him/herself from the
responsibility for what is being reported" (Martin & White 2005: 113). Obviously, this
maximises the space for dialogistic alternatives. Example (42) is a constructed example,
since the corpus does not contain an example of distancing. An explanation for this will be
given in the following section.
4.1.3.2 Summary and interpretation
The outsourcing or acknowledgement of responsibility on behalf of the speaker is thus
more complicated than initially thought. The main engagement types to express
responsibility are represented on the axis below, from low to high responsibility
acknowledged by the speaker.
56
FIGURE 3. Responsibility for the statement expressed by the speaker
Because of the dialogistic approach, the types also take into account the response by
a putative dialogic partner. The possibility of dialogic alternatives is represented by Figure
4, from contractive, i.e. excluding dialogic alternatives, to expansive, i.e. invoking dialogic
alternatives.
FIGURE 4. From contraction to expansion
As seen in the previous section, (almost) all engagement subcategories make use of
evidential expressions. Now it is important to discover which engagement options, which
are seen as functional motivations, are at work in the popularised form of science, and are
consequently responsible for the division of the different evidentiality types in the corpus.
To find this, we need to begin at the starting point: the genre.
The documentary makers work with a narrator who talks most of the time. One may
say that he is moved by two communicative purposes. On the one hand, it is his job as an
objective narrator to stay in the background and to explain things or to report what other
people say. Objectivity corresponds to neutrality. So, when looking at figure 3, this first goal
is reminiscent of acknowledgement. When applying the category 'acknowledgement', the
narrator acts indeed as an 'informational fair trader'.
On the other hand, the narrator has to bear in mind the aim of the documentary,
which is explaining scientific facts to laypeople and convincing them of certain scientific
findings. For this task he needs undeniable facts and a low input of contradictory opinions.
The combination of Figure 3 and Figure 4 shows that endorsement corresponds to these
needs. The propositions are presented as true and convincing, because the narrator himself
Distancing Acknowledgement Endorsement
Low responsibility High responsibility
Contraction Expansion
Denial Concurrence Entertain Acknowledgement
Pronouncement Distancing
Endorsement
Expansion
57
takes the responsibility for it, which means he trusts the source, and therefore not allows
contractive alternatives.
The specific aim of the genre is also responsible for the small role of distancing. The
narrator and the scientists try to convince the audience, not to distract them or make them
doubt. In addition, because of the important role of the narrator, the categories
concurrence and pronouncement are of a minor importance in the documentary. They
imply the intervention of a person, and therefore they will only influence the scientists who
feature in the documentary. This is also the reason why, next to the modality adverbials
and the suggestions or beliefs expressed by scientists, the representation of entertain
examples will be scarce in the documentary. Finally, denial and countering are of no
importance because they do not invoke an evidentiality structure, unless in the subtitles.
In conclusion, when starting from the needs and aim of the genre, endorsement and
acknowledgement are the main motivations to use evidentials in a documentary. This
observation explains the high incidence of reported evidence (people say that) and
inferential evidence (research shows) found in the corpus.
58
4.2 English speech versus Dutch subtitles
This comparative study focuses on the subtitles of the English evidentiality expressions. Via
the comparison between the Dutch subtitles and the English original sentences, the present
part aims to shed light on the multifunctionality of the English evidential markers.
Furthermore, the investigation of the translations offers an insight into the evidentiality
system of Dutch and the motives of the subtitler. First of all, section 4.2.1 provides the
differences between original and translated text in terms of expressions of evidentiality.
The section is divided into two parts, which outline the distinctions between the English
and Dutch evidentiality types on the one hand, and the differences in formal encoding on
the other hand. Subsequently, the reason for the drastic differences is searched in the
subtitling restrictions (section 4.2.2.) and finally, it is examined whether the differences
may be due to different functional motivations of the subtitler.
4.2.1 Differences
A slight majority (53,70%) of all the Dutch subtitles offers a specific and literal translation of
the English evidentiality expressions, as can be seen in example (1):
(1) The repeated ultra precise measurements reveal that land here on the
surface hardly moves at all.
De herhaalde superprecieze metingen onthullen dat de grond aan het oppervlak
amper beweegt.
As a result, the following Dutch evidentiality verbs are derived from the translations. Direct
evidence, which is indicated by English verbs such as to show, to notice and to reveal, is
expressed in Dutch by the verbs (aan)tonen, laten zien, (op)merken and onthullen. Secondly, the
verbs of saying which mark reported evidence are translated into Dutch as: vragen (to ask),
begrijpen (to understand), bevestigen (to confirm), geloven (to believe), zeggen (to say), voorstellen
(to propose), waarschuwen (to warn). The Dutch verbs that indicate inference are: bewijzen (to
prove), bewijs leveren (reveal evidence), aannemen (to assume), moeten (must) and aangeven (to
indicate). Lastly, the Dutch verbs weten (to know), denken (to think, to have a hunch) and
vermoeden (to suspect) express certainty or uncertainty.
All the literal translations are characterised by an evidentiality verb. The verbs
mentioned above are thus the recurring Dutch evidentiality verbs in the science
documentary. In comparison with the evidentiality forms proposed by de Haan (modal verb
59
moeten, modal verb zouden, raising verb schijnen and the perception verbs), the current
corpus provides a broader view on the Dutch evidentiality verbs.
Beside the literal translations, 118 cases of the 255 tokens (46,27%) show a difference
between the original English evidentiality expression and its Dutch subtitle. These
differences range from minute alterations, like a distinct verbal expression, to more
significant ones, like the change of evidentiality type. Accordingly, the categorisation that
follows encompasses two classes. The first one (section 4.2.1.1) discusses the distinctions
with regard to the types of evidentiality, i.e. either the type is changed or not expressed in
Dutch, while the second one (section 4.2.1.2) deals with the formal possibilities of
evidentiality in English and in Dutch.
4.2.1.1 Types
This section starts with examples of the most drastic change found among the 118 subtitles
which differentiate from the English evidentiality expressions. The subtitles of ten English
sentences demonstrate a complete deletion of the evidential meaning by reducing the
English evidentiality expression to a pure declarative statement. This omission occurred
with direct evidence types, as well as with reported and inferential evidence types.
Examples (2) to (4) illustrate the removal of the underlined evidentiality markers for the
three types:
(2) We see that the secret of the slipping San Andreas fault is actually the rock
itself. (Direct evidence)
Het schuiven van de San Andreasbreuk wordt veroorzaakt door de steen zelf.
(3) People agree that only a massive force could have created such a huge crater.
(Reported evidence)
Zo'n krater kon alleen door een enorme kracht zijn veroorzaakt.
(4) It turns out there's a really strong relationship between the age of the
seafloor and its depth in the water. (Inferential evidence)
Er is een sterk verband tussen bodemouderdom en waterdiepte.
Another example, in contrast with the data above, is the unique occurrence that is marked
by the addition of an evidential meaning. The English source sentence does not express
evidentiality, but the subtitler adds a hearsay evidence marker (men denkt) which signals
uncertainty.
60
(5) A tsunami isn't over in just a few seconds.
Men denkt dat een tsunami zo voorbij is.
The reason for the addition or deletion of the evidential meaning in the subtitles will be
discussed in section 4.2.2 and 4.2.3, where the subtitles are analysed more closely.
Next to the deletion or addition of an evidential meaning, 52 Dutch subtitles contrast
clearly with the original expressions because they belong to a different evidentiality type.
The numbers are represented in Table 3.
TABLE 3. Change of types from English to Dutch
English/Dutch Direct Reported Inferential (Un)certainty
Direct 19
Reported 25
Inferential 2
(Un)certainty 5 1
25 evidentiality expressions change from the English reported evidence type (to believe) to
the Dutch expression of uncertainty (denken). 19 examples indicate in English direct
evidence (to show, to reveal) whereas their subtitles indicate inferential evidence (bewijzen,
blijken). The English expression of (un)certainty (to know, to think) is represented by six
tokens, five of which transfer to reported evidence (volgens), one to inference (moeten) and
two go the other way around, i.e. from inference (to suggest) to uncertainty (denken).
The largest group (25 occurrences) establishes a change from the reported evidence
type (R) to the expression of (un)certainty (U/C). 19 occurrences of this group are
characterised by the verb to believe. Originally, to believe was considered a marker of
reported evidence, whereas in Dutch it is 19 times translated as the expression of
uncertainty with the verb denken ('to think'). What is more, in the complete corpus only six
subtitles translate to believe as indicating reported evidence. This is obviously due to the
double meaning of the verb to believe, which is first 'accept that (something) is true,
especially without proof' and second 'hold (something) as an opinion; think'.5The
comparison below illustrates the two meanings respectively:
5 Definitions are quoted from the Oxford dictionary, consulted online: http://oxforddictionaries.com/definition/english/believe?q=believe, 14/05/2013
61
(6) Lyell believed the world wasn't shaped in a few days or even years. (R)
Lyell geloofde dat de wereld niet in enkele dagen geschapen was. (R)
(7) Lyell believed this process had been going on for thousands of years. (R)
Lyell dacht dat dat proces al duizenden jaren gaande was. (U)
In a similar way, the second largest group (19 occurrences) deals with the verbs to show and
to reveal. Even though I have treated to show as a direct evidence marker because of its
causative perceptive character explained in section 4.1.1.1, some subtitles (9 occurrences)
translate to show as indicating inference by using the verbs: uitwijzen, bewijzen, blijken.
Example (8) illustrates the change from direct (D) to inferential evidence (I).
(8) Extinct volcanoes show that this process began on the coast. (D)
Dode vulkanen bewijzen dat dit proces aan de kust begon. (I)
In comparison, only ten tokens of the whole corpus contain subtitles which translate to show
as a direct evidence marker with the verbs (aan)tonen and laten zien.
Similarly, the verb to reveal was also treated as indicating direct evidence, but is
sometimes translated as indicating inference. Of the 24 direct evidence examples with to
reveal, ten express inference in Dutch with verbs such as bewijzen ('to prove'), erop duiden
('to indicate') and blijken ('to become clear'). Example (9) shows a translation that indicates
direct evidence, whereas in examples (10) and (11) the subtitles indicate inference. The
translation with blijken will be discussed in the following section. When to reveal occurs in
combination with the noun evidence, it is considered as indicating inference in English as
well as in Dutch, as illustrated by example (12):
(9) Doctor Roberts' data reveal that this whole glacier is sliding forward at an
astounding rate of two feet per day. (D)
De metingen van dokter Roberts laten zien dat deze gletsjer met een
verbazingwekkende snelheid van 60 cm per dag verschuift. (D)
(10) The pumice from 1883 reveals that the Krakatau was a stratovulcano. (D)
Het puimsteen uit 1883 bewijst dat de Krakatau een stratovulkaan was. (I)
(11) Microscopic analysis reveals the grains have been cemented together by
chemicals and pressure. (D)
Uit analyse blijkt dat de korrels samengeklonterd zijn door chemicaliën en
druk. (I)
62
(12) The nest reveals strong evidence that plants and mammals did exist here. (I)
Het nest bewijst dat hier vroeger planten en zoogdieren leefden. (I)
These observations, which are based on the subtitles, prove that, in contrast with the
assumptions expressed in the first study, the three English verbs to believe, to show and to
reveal are able to express more than one type of evidence.
After having mentioned the translation of English reported evidence into the Dutch
expression of (un)certainty, this paragraph deals with the opposite. Examples include:
(13) Merguerian is convinced that the same thing happened in Central Park. (C)
Volgens Merguerian is dit ook in Central Park gebeurd. (R)
(14) Tankersley is convinced catastrophe drove the mammoths to extinction. (C)
Volgens Tankersley zijn de mammoeten door een ramp uitgestorven. (R)
(15) Geologist David Franzi knows that only a raging torrent could have shifted
them. (C)
Volgens David Franzi kan alleen een woeste stroom ze hebben verplaatst. (R)
Example (13) is analogous with example (24) discussed in section 4.1.1.4 of the first study
and here recapitulated as example (14). When discussing example (14) in the first analysis,
the pragmatic inference technique was applied. It was said that, next to its expression of
certainty, one may interpret the sentence as indicating direct evidence (the narrator
deduced the statement from Tankersley's convincing behaviour), reported evidence
(Tankersley himself told it to the speaker), as well as inferential evidence (the narrator
inferred this conviction from research by Tankersley). As a result, instead of the literal
translation Tankersley is ervan overtuigd dat, the subtitler decides to ignore the certainty
expression and translates the sentence with the structure volgens X ('according to X'), which
denotes that the statement is derived from something X said (reported evidence). Similarly,
in example (13) and (15), the subtitler preferred to ignore the certainty expression and
indicates, in these cases, quotative reported evidence. This and other uses of the
preposition volgens will be examined concisely in the section that follows.
Finally, in example (16) the subtitles add an evidential type to the original one:
(16) These bubbles suggest to geologists that the Krakatau was a stratovulcano. (I)
Hierdoor denken geologen dat de Krakatau een stratovulkaan was. (U/I)
63
At first sight, example (16) expresses a change from inference (these bubbles suggest) to
uncertainty (geologen denken), but the inferential meaning still remains present through the
word hierdoor ('because of this').
In sum, one notices a variety of differences in type between the English and the
Dutch evidentiality expressions. Nevertheless, the use of these particular subtitles are
justified by the multiple evidential meanings of certain English verbs, the application of the
pragmatic inference technique or the addition of an epistemic marker beside the original
evidentiality type.
4.2.1.2 Formal encoding in Dutch
After having treated the semantic differences, the current section zooms in on the formal
differences between the English evidentiality expressions and their Dutch subtitles. The
data which represent a change in type are re-evaluated in terms of their formal encoding.
This analysis is followed by the subtitles which differ in formulation but indicate the same
type of evidence.
The arguments of the 'difference in type' examples mentioned in the previous
section are characterised by a change of function. For the formal analysis, additional
examples, which represent the deletion of the evidential meaning, the addition, and
the change of type respectively are given:
(17) He noticed below the falls was a great gorge which locals said was steadily
increasing in length as the water wore away the edge of the falls.
Onder de waterval lag een grote kloof die steeds langer werd naarmate de
rand van de waterval wegsleet.
(18) A tsunami isn't over in just a few seconds.
Men denkt dat een tsunami zo voorbij is.
(19) These bubbles suggest to geologists that the Krakatoa was a stratovulcano.
Hierdoor denken geologen dat de Krakatau een stratovulkaan was.
The change of type as a translation strategy is represented by example (16), here re-
numbered as (19), which was described as adding uncertainty to the inferential type. The
other changes (from reported to uncertainty, direct to inference etc.) given in the previous
section, are not useful for this formal analysis. The structure of these subtitles does not
64
differ from the source structure, except when the translations use blijken and volgens which
will be discussed below as separate categories.
When analysing the examples above, which do express a formal discrepancy, one
notices the following. First of all, example (17) shows that the divalent verb to notice has a
human subject (he) and a that-clause as object (even though that is not expressed). In Dutch
the that-clause becomes the main clause, because the evidential marker (he noticed) is
deleted. As a result, the internal structure of the that-clause is maintained, and the subject
(a great gorge) of the that-clause becomes the subject (een grote kloof) of the Dutch main
clause. The other examples (cf. section 4.2.1.1) with a deletion of the evidential meaning are
translated according to the same principle. Secondly, in the example of addition, the
opposite occurs. The main English statement becomes a dat-clause in Dutch. Thirdly, in
example (19) the change of type causes a change of arguments. In the original sentence the
subject denotes a nonhuman thing (these bubbles), and the object is a that-clause. In the
translation, by contrast, the subject is human but the object remains a dat-clause. What is
striking is that geologists is a prepositional object with a benefactive function in the source
text while in the Dutch translation geologen functions as the subject. So, even though the
two verbs are divalent, and using a that-clause (I suggest that, Ik denk dat), the arguments are
changed.
Concerning the formal encoding of the data without change in type, four different
Dutch formulations are distinguished. First one finds a small amount of the preposition
volgens (according to) (11 occurrences), followed by 17 occurrences of the verb blijken (to
become clear), 24 of men (one/they) and finally, 26 examples of various other structures.
First of all, the smallest group is typified by the preposition volgens. Next to the
examples in section 4.2.1.1 (cf. examples 13, 14 and 15) which express certainty, seven
tokens of which the subtitles are characterised by volgens express reported hearsay
evidence (cf. example 20) and one signals inferential evidence (cf. example 21). These
examples prove that the preposition volgens is used to indicate either that the statement is
derived from something a person said (reported evidence) or that it is derived from actual
facts (inferential evidence), although the latter occurs less frequently in the corpus.
(20) Experts believe that in the US New York is the third most vulnerable city
after Miami and New Orleans, for a hurricane disaster.
Volgens experts is New York de op twee na kwetsbaarste Amerikaanse stad na Miami
en New Orleans, voor een orkaanramp.
65
(21) The study estimates that a major earthquake in the LA metro area would cost
2000 dead, 50.000 injuries and 200 billion dollars of damage.
Volgens een studie uit 2008 zou een zware beving in LA 2000 doden, 50.000 gewonden
en 200 miljard dollar schade eisen.
Most of the subtitles with volgens change the NP + verb structure or the PN + verb group. The
subtitler turns the English human subject X into a part of the prepositional phrase volgens X.
As can be seen in example (20), the subject which acts in the English sentence (experts) is
preceded by volgens in the Dutch subtitles. Again, the object, which is a that-clause, becomes
the main clause. Consequently, the subject of the that-clause (New York) becomes the subject
of the main clause. This causes the deletion of the original verb in favour of the verb of the
that-clause. Semantically speaking, the agent of the original sentence (experts) becomes a
source in the subtitle. This is also the case with the sentences expressed above, which
express certainty.
Nevertheless, in spite of the frequent use of the preposition volgens, 31 English
expressions with a human noun or proper noun as subject are subtitled literally, of which
example (22) is an illustration:
(22) Geologists believe that the trench formed the volcanoes by a process called
subduction.
Geologen denken dat de vulkanen ontstonden bij een proces dat subductie heet.
Aside from the literal translation and the structure volgens X, the reported evidence is also
introduced by the pronoun men, which occurs in 24 subtitles. Men is used with expressions
of reported evidence (10 occurrences), (un)certainty (10 occurrences) and inference (4
occurrences).
First, when looking at the reported evidence type and the (un)certainty type, the
generic subjects scientists, geologists, etc. and the personal pronoun they are replaced by the
more general term men. As deducible from example (23), scientists now believe is translated
more generally as men denkt, which justifies the classification of the generic nouns as
subjects of hearsay reported evidence instead of quotative. Furthermore, the replacement
of the subject by men does not affect the internal structure, given that men is also subject in
the Dutch sentences. In addition, its use reinforces the statement made in the first study,
which said that for the target group of the science documentary the fact that the
information being communicated is the result of scientific research is more important than
66
knowing exactly who carried out this research. Additionally, men is the default translation
of the passive voice (cf. example 24).
(23) Many scientists now believe that the damage was greater than expected
because the 1906 had traveled a super shear speed.
Men denkt nu dat de schade groter was dan verwacht doordat de aardbeving uit 1906
een supershear snelheid had.
(24) It is thought that eruptions began in this area over 36 million years ago.
Men denkt dat de uitbarstingen hier 36 miljoen jaar geleden begonnen.
Secondly, four cases of inference are translated by men. At first sight, this is surprising
because most of the inferential subjects are nonhuman (research, clues, dating, etc.), but two
of these four cases are assumptions, in which humans are implied. The two other
expressions follow sentence (25), in which a human subject is introduced by men:
(25) Ice cores From Antarctica suggest that Krakatau may have exploded before.
Toch leest men uit ijskernen af dat hij al eens uitgebarsten kan zijn.
A third group is characterised by the verb blijken. 12 of these 17 sentences express
inference, followed by five examples of direct evidence with the verb to show which are
mentioned in the previous section because of their change of type. Example (26) and (27)
exemplify the two possible structures with blijken. In example (26) the subject of the
inferential sentence, that expresses the crucial evidence Y from which the statement is
inferred, is changed into the pattern uit/na Y blijkt + statement. Similar to the volgens group,
the subject is changed into an adverbial.
(26) Radiometric dating proves this rock is about 450 million years old.
Uit radiometrische datering blijkt dat de steen 450 miljoen jaar oud is.
(27) The map revealed that the Mariana trench is just a tiny fraction of a network
of enormous underwater canyons, stretching right around the planet
De Marianentrog bleek te behoren tot een netwerk van enorme zeekloven dat de hele
planeet omvatte.
Example (27), by contrast, is an instance of subject to subject raising (Auwera & Noël 2011).
Instead of saying Het bleek dat de Marianentrog behoort tot een netwerk van enorme zeekloven dat
de hele planeet omvatte, the subject of the dat-clause is raised to the subject of the main
clause, followed by the verb blijken and a te-infinitive. In a similar way, when comparing the
67
English and Dutch sentences (cf. example 17,20,21,26,27), the Dutch translation offers a kind
of 'subject to subject raising' as well. The subject of the that-clause in English is raised to the
subject of the main clause in the subtitles.
In addition, the use of blijken causes the omission of the expert in direct evidence
examples, such as
(28) Recreating how the land had moved showed Hudnut that the two parts of the
creek had traveled more than three hundred feet apart.
Uit reconstructie van de beweging van het land blijkt dat de twee delen van
beek ruim 90m uit elkaar geschoven zijn.
In example (28) one notices a decrease in valency because of the use of the verb blijken. The
subject (recreating how the land had moved) is followed by the predicate (show), the indirect
object, and the object (that-clause). The Dutch translation, on the other hand, consists of two
arguments, one being a prepositional phrase as counterpart of the English subject, the other
being the dat-clause, which is the translation of the that-clause. The human beneficiary is
deleted because of blijken.
The deletion of an argument happens also with the verb schijnen. In example (29) the
reporters are deleted, due to the use of the specific Dutch verb. Schijnen designates that the
utterance is based on what is said.6 As a consequence, schijnen encompasses perfectly the
hearsay reported evidence which is expressed by people say.
(29) People say that more than half a million people died from the drought in
India.
Daarbij schijnen meer dan een half miljoen mensen te zijn omgekomen.
The last group comprises various forms. First of all, the Dutch subtitles tend to verbalise
the English expressions. For instance, to be proof becomes bewijzen, to be clues becomes
vertellen, it is a clue becomes erop wijzen/erop duiden, to be evidence becomes bewijzen, to come to
the conclusion becomes concluderen. What is more, the expression to be proof is never
translated literally in the complete corpus, while to indicate becomes een aanwijzing zijn,
evidence suggests becomes er is bewijs dat and to prove becomes bewijs hebben in only three
occurrences. Furthermore, in spite of the frequent use of volgens, the English counterpart
according to is never used.
6 Translation of definition found in Van Dale dictionary, consulted online http://www.vandale.be/opzoeken?pattern=schijnen&lang=nn#.UZONncoyc80. 05/05/2013
68
As with the blijken group, some subtitles reduce the evidential markers to one
element. The Dutch subtitles of example (30) express inference only by the conjunction dus,
while in example (31) the additional reported evidence leading investigators to conclude
disappears and the inferential meaning remains:
(30) Finding this told them the Mid-Atlantic ridge was highly volcanic.
De Mid-Atlantische Rug was dus zeer vulkanisch.
(31) There are hundreds of layers of salt, leading investigators to conclude the sea
must have dried up and refilled hundreds of times.
Er zijn honderden zoutlagen. De zee moet dus honderden malen drooggevallen en weer
volgelopen zijn.
Other distinct formulations are the use of an adverb (vermoedelijk) to replace a NP + Verb
structure (scientists suggest), a lexical phrase (naar we denken) instead of a hearsay translation
(it is thought), the transfer from direct to indirect speech, and the deletion of a conjunction
like so.
To sum up, the Dutch language uses some particular structures to express
evidentiality. First, the pronoun men illustrates clearly the unspecified nature of the
hearsay subjects, but does not change the internal structure drastically. Second, the
preposition volgens is used to express reported evidence. When volgens is used, the English
subject, which is the agent, turns into an adverbial which expresses the source. This change
from subject to adverbial also happens with the first possibility of the verb blijken, which is
uit/na Y blijkt + statement, where Y denotes the inferential evidence. The second possibility
with blijken is subject to subject raising, which means that the subject of the that-clause
becomes the subject of the main clause. This raising can be applied to the subtitles in
general. Most of the examples of the volgens group and the blijken group are translated as
follows. The subject of the English that-clause becomes the subject of the Dutch main clause,
which could be considered as the subject of the that-clause that 'raises' to the subject
position in the Dutch translation.
4.2.2 Influence of subtitles
The radical differences in formulation, as expressed above, are often caused by the
subtitling restrictions, led by reading speed and synchronicity. This section focuses on the
69
shortened subtitle sentences in order to find out whether (part of) the evidentiality
meaning is deleted because of the advantage the shortening offers for the viewer.
In the literature review, I introduced the following subtitling techniques: (1)
condensing the text, (2) omission or paraphrase, (3) muddled speech, (4) ellipsis, (5)
merging short dialogues, (6) simplifying the syntax, (7) simple vocabulary, (8) subtitle
breaks, (9) borrowing time. I will not focus on (3) muddled speech and (5) merging short
dialogues. The use of dialogues in a television documentary is minimal, or in this case,
nonexistent because the majority of the text is spoken by means of a voice-over. The other
parts consist of monologues by professors, geologists or other scientists. They speak as
clearly as possible to explain their theories or ideas. Consequently, in the current corpus,
neither muddled speech nor dialogues occur.
The remaining strategies can be divided into three categories, being deletion, time
or simplifying strategies. As seen previously, some subtitles delete (evidential) parts of the
original English sentence. The condensation of the sentences is due to the limited time
and/or space to display the subtitles. Moreover, the restrictions on time and space are led
by the reading speed of the viewers and the need of synchronicity between subtitle and
sound and subtitle and image. According to Ivarsson and Carroll (1998) the reading speed
is much slower than for instance when reading a book, because of the images and sound the
viewer has to absorb while reading the subtitles. For a subtitle to be completely read and
understood, one line should be kept on the screen for about three seconds, while one and a
half lines need at least four seconds (Ivarsson & Carroll 1998). In addition, wwhen dealing
with reading speed, it is important to bear in mind the target group. For this documentary,
which is originally broadcasted on History Channel, one has to take into account all kinds of
audiences; youngsters who are accustomed to subtitles and the English language, as well as
elders who experience more difficulties with the multimodal genre. The mix of both
possible target groups automatically slows down the reading speed.
Next to reading speed, another important notion to remember is the synchronicity
between the subtitle and both the image and the sound. The most important thing for the
subtitler is the beginning and end of the subtitles which have to coincide with the speech of
the speaker. In addition, it is also crucial that the subject of which the narrator is talking, is
displayed on the screen. (Ivarsson & Carroll 1998).
The following table, which is based on de Linde and Kay's (1999) analysis, sheds light
on the reading speed and synchronicity of some data.
70
TABLE 4. Time analysis of a 'deletion' example
On-set Off-set
Time in seconds 0 seconds (01:39) 5 seconds (01:44)
Speech People agree that only a massive force could have created such a huge crater.
Time in seconds 0 seconds (01:39) 5 seconds (01:44)
Subtitle Zo'n krater kon alleen door een enorme kracht zijn veroorzaakt.
Time in seconds 0 seconds (01:39) 5 seconds (01:44)
Image Image of the crater
Table 4 illustrates how long the speaker needs to utter the sentence (5 seconds), how long
the subtitle appears on the screen (5 seconds) and how long the image remains the same (5
seconds). First of all, according to the optimal time in seconds proposed by Ivarsson and
Carroll, the subtitle which consists of two lines appears long enough on the screen to be
read and understood by the viewer. Secondly, the 'on-set' and 'off-set' times of the speech
coincide with those of the subtitle. In the corpus in general, a subtitle never appears longer
on the screen than the voice of the speaker, even though two seconds without speech
follow. Thirdly, the image does not change during the speech and the appearance of the
subtitle. As a result, the criteria of synchronicity and reading speed are fulfilled. This would
not have been the case when the English expression would have been translated literally:
Men is akkoord dat zo'n krater alleen door een enorme kracht kon zijn veroorzaakt. The non-
abbreviated sentence needs more than two lines, and therefore requires more seconds to be
read and understood, which would challenge the synchronicity between the speech and the
subtitle. Also the synchronicity between the subtitle and image would change, since a new
line commonly coincides with a new image.
When a subtitle translation does consist of three lines, a good application of subtitle
breaks is needed. In general, the subtitle breaks (//) are marked in accordance with the
intonation of the speaker or with the image that follows. The division of the example below,
which, albeit the deletion of the evidential meaning, consists of three lines, is done as
follows:
71
TABLE 5. Time analysis of a 'deletion' example
On-set Off-set
Time in seconds 0 seconds (17:30) 7 seconds (17:37)// 10 seconds (17:40)
Speech The map revealed that the Mariana trench is just a tiny fraction of a network of
enormous underwater canyons //
stretching right around the planet
Time in seconds 0 seconds (17:30) 7 seconds (17:37)// 10 seconds (17:40)
Subtitle De Marianentrog bleek te behoren
tot een netwerk van enorme zeekloven //
dat de hele planeet omvatte.
Time in seconds 0 seconds (17:30) 7 seconds (17:37) 10 seconds (17:40)
Image Image of the earth from
above
Close-up of the earth
In this case, the speaker pauses after the words underwater canyons, which means the long
first part (cf. image below) has to be read and understood during seven seconds. In addition,
the part that follows the underwater canyons coincides with a new image, which means the
first part has to be read before the image changes. This is important because "when a shot
change occurs in the middle of a subtitle, viewers return to the beginning of a partially read
subtitle and start re-reading" (Baker 1982 qtd. in de Linde & Kay 1999: 48). Accordingly, the
subtitler deletes the first words, in order to be sure the first long part of the sentence is
read and understood before the image changes. Additionally, the sentence has to be
shortened because the first part already occupies the available space:
72
The synchronicity between subtitle and image has also an influence on the reading
speed. In the example below, Dr. Cheryl Jaworowski of the Yellowstone national Park utters
the first sentence in five seconds, so the first image, which portrays Dr. Cheryl herself, and
the concomitant subtitle appears for five seconds. She talks fast and the second sentence
follows immediately after the first. What is more, the second sentence is accompanied by a
new image, which displays the hills and trees she was talking about. The viewer needs a
split second to absorb the new image and to start reading again. So, in order to take into
account the time the viewer will need to read the two sentences and to synchronise with
the speech and image, the first words of the second sentence are left out.
(32) Today, when you're looking out across the landscape, you see here hills and
trees. //
Nu zie je hier alleen glooiende heuvels en bomen. //
So, obviously, there's not a big hole left in the ground from the eruption of
the Yellowstone volcano.
Er is geen gat van de uitbarsting van de Yellowstone-vulkaan.
These three examples show that, sometimes, the subtitler has to condense the source
sentence to meet up with the reading speed and synchronicity, so that the viewer is able to
read and understand the subtitles properly. Nevertheless, this does not mean the viewer
misses crucial information. The motivation to delete particular parts is discussed in the
following section.
Finally, the last technique is the simplification of the syntax and vocabulary. As
stated in the literature review, using simple vocabulary is only done when it does not affect
the style and spirit of the genre (Ivarsson & Carroll 1998). No data were found in which
geographical or scientific difficult terms were simplified in the subtitles. What is more, the
complicated terms used by the speaker are explained in the documentary itself. Examples
include:
He proposed that Hawaii was created by something called a 'hotspot', an
exceptionally hot region beneath the earth's crust, that was
concentrated under Hawaii's Big Island.
73
Concerning the syntax, the most frequent distinctions in sentence structure were found
with the structures volgens X and uit/na Y blijkt + statement. Even though they may cause a
change in arguments, they are not considered as simplifying the syntax, given that the
original syntax is not difficult at all.
4.2.3 Functional motivations of the subtitler
As mentioned in the literature review, when the original sentence has to be condensed the
subtitler has to decide what is the essential content. This decision is affected by the aim of
the genre, which consists of providing a vital explanation to people outside of the scientific
community. Consequently, the essential information in a documentary is the statement
which contains the information and/or explanation needed to understand the message.
Therefore, when condensation is required, the need to express the fundamental
information excludes the accessory matters, which are in the 'deletion' examples, the
sources of information.
Following Martin and White's forms of engagement (denial, countering,
concurrence, pronouncement, endorsement, entertain, acknowledgment and distancing)
examples (33) to (35) represent acknowledgement, endorsement and concurrence
respectively. The speaker of example (33) is presenting the statement (cf. underlined part)
as a neutral observer. Nevertheless, by saying people agree he acknowledges that several
persons agreed upon the statement, which increases the degree of certainty. Example (34)
represents endorsement, which means the speaker sees the statement as correct, or
undeniable. The example of concurrence (cf. example 35) expresses, via the certainty
adverb, an agreement with the addressee. Thus, when looking at examples (33) to (35), it is
noted that the evidentiality expressions imply a high degree of certainty. This certainty is
also expressed by the bare assertion in the translation. So, from an epistemic point of view,
the English expressions and their translations express the same meaning. Both languages
show that the statements are facts. The evidence for the fact (a map) or who reported them
(people) becomes accessory, since the epistemic meaning is considered as more essential
than the evidential one.
(33) People agree that only a massive force could have created such a huge crater.
Zo'n krater kon alleen door een enorme kracht zijn veroorzaakt.
74
(34) The map revealed that the Mariana trench is just a tiny fraction of a network
of enormous underwater canyons stretching right around the planet.
De Marianentrog bleek te behoren tot een netwerk van enorme zeekloven dat
de hele planeet omvatte.
(35) Today, you see here hills and trees. So, obviously, there's not a big hole left in
the ground from the eruption of the Yellowstone volcano.
Nu zie je hier alleen glooiende heuvels en bomen. Er is geen gat van de
uitbarsting van de Yellowstone-vulkaan.
In addition, the examples above illustrate that not only the epistemic meaning remains the
same, but also the motivations to use a particular structure are the same for the subtitler as
for the English speaker. If possible, the subtitler aims to maintain the functional motivation
that moved the speaker to use an evidential marker in the source language. The
endorsement examples (36) and (37) show how the literal translation expresses the same
certainty and validity added by the speaker as the non-literal translation where the fact
that there is clear evidence is implied in the verb blijken (to become clear). In example (38) as
well, in spite of the different structure, the specific meaning of schijnen (according to what
they say) expresses the neutral reporting meant by the English speaker.
(36) Contemporary reports show that most of Krakatau victims were not killed by
these deadly avalanches.
Recent onderzoek wijst uit dat de meeste slachtoffers niet door gloedwolken
zijn omgekomen.
(37) This research shows that although the air in the desert is around 10 %
humidity, on rare occasions it rises as high as 75 %
Het bleek dat de luchtvochtigheid die hier gemiddeld 10 % is bij zeldzame
gelegenheden kon oplopen tot 75 %.
(38) People say that more than half a million people died from the drought in
India.
Daarbij schijnen meer dan een half miljoen mensen te zijn omgekomen.
The most significant example, which illustrates that the subtitler aims at expressing
the same meaning as the source sentence, is the unique example of denial.
75
(39) A tsunami isn't over in just a few seconds.
Men denkt dat een tsunami zo voorbij is.
As seen in section 4.1.3, the English denial is used to anticipate beliefs by others and to
refute them. This particular meaning is expressed explicitly in Dutch by the use of the
evidential markers men denkt ('people think'). So, the case of denial proves that, albeit the
different structure, the subtitler and the speaker share the same functional motivation,
which is in this case denying the general beliefs.
76
5 Discussion
The present chapter discusses the findings of the corpus analysis. In particular, it zooms in
on three aspects treated in the analysis:
1. The evidentiality types
2. The formal encoding of evidentiality
3. The functional motivations for evidentiality
All three topics examined the English expressions and their accompanying subtitles. The
results of these two studies will be combined and summarised in the sections that follow.
When interpreting the results, it is important to bear in mind that this is a small-
scale corpus, limited to the specific language of a science documentary. This popularised
genre of science and its particular aim has a considerable influence on the results.
5.1 Types of evidentiality in popularised science
In the first study four different types to express English evidentiality were distinguished.
These distinctions were based on the findings presented in the literature review and consist
of the indication of direct evidence, reported evidence, and inferential evidence and the
expression of (un)certainty. The same types were examined when dealing with the Dutch
subtitles. This comparative approach helped to gain more insight into the features of the
four evidentiality types found in the popularised genre of science, i.e. the television
documentary.
The results concerning the direct evidence type suggest that, albeit the
assumptions in the literature review, in this particular corpus the pure perception verbs are
not the main indicators of direct evidence. The majority of the data indicates direct
evidence with the causative perception verbs to show and to reveal. Nonetheless, these verbs
are not exclusively used for the direct evidence type. The subtitle analysis revealed that the
two verbs may also indicate inference, which shows the evidential multifunctionality of the
English verbs.
In a similar way, with regard to the reported evidence, the occurrences of to believe
were systematically treated as reported evidence, but the translation warned that its
meaning should be analysed separately and based on the context. The translation study also
helped distinguishing between hearsay and quotative evidentials. The subjects of the
77
reported evidence, i.e. generic nouns like scientists, geologists etc., are considered hearsay
subjects. This decision is backed up by the translation of these generic subjects, which is in
most cases the general Dutch term men. The frequent occurrence of generic hearsay
subjects resulted in a high incidence of hearsay expressions, which is due to the specific aim
of the popularised scientific discourse. For laypeople, it is more important to know that the
claims are based on actual research than to hear the names of the (for them unknown)
scholars behind it. The slightly lower number of quotative expressions may be explained by
the fact that the narrator only names the speaker who will feature in the documentary.
Thirdly, the low number of quotation marks is due to the specific multimodal genre.
Because the speaker is shown on the screen and his/her name is displayed, quotation marks
are not needed. Also, in a scientific narrative it is easier to use indirect speech.
Another consequence of the genre-specific study is the adjustment of the
inferential evidence type. The inclusion of sentences like example (1) as inferentials was
justified by tailoring the analytical toolkit to the genre under investigation, which is
centred around searching for and providing evidence.
(1) The crystals are clear evidence that under the springs is a volcano.
Firstly, this reconsideration has as result that the subjects of the inferential evidence
expressions are all nonhuman. However, personified subjects do also occur. The
investigation proves that the examples of personification have to be classified according to
the level of humanity (reported evidence) or investigation (inferential evidence) implied in
their source. Second, the results show that the verbs on which the recognition of evidential
expressions in the science documentary was based (cf. Figure 2), did not appear in this
genre. Ifantidou's verbs were derived from spontaneous conversations, where the direct
speaker-hearer interaction is central, but this interaction does not take place between
viewer and narrator. The different communicative needs of the two genres result in
different evidential verbs and prove that a genre-specific approach to evidentiality is
needed.
Even though Ifantidou's verbs are not represented, her inclusion of the expression
of (un)certainty as a part of evidentiality, is. Again, the (structure of the) documentary
itself is responsible for the occurrence of the (un)certainty expressions. As explained in the
data collection (section 3.1), the documentary starts from questions and doubts and ends
78
with evidence and conclusions, which creates expressions of both certainty and
uncertainty. The findings reinforced the statement that in non-grammatical evidentiality
languages the expression of evidentiality frequently coincides with epistemic expressions.
Moreover, in the translation as well, the addition of an epistemic marker beside the original
English evidentiality type was noticed. Lastly, also the use of modal must as an evidential
marker is proved by the data.
5.2 The formal encoding of evidentiality in popularised science
One observes in the results that 213 of the 255 English tokens mark evidentiality by means
of the following formations: noun phrase (NP) + verb, proper noun (PN) + verb and pronoun (Pron)
+ verb. The high frequency (83,53%) of the structure subject + verb + that-clause contrasts with
the variety of forms which exist to express evidentiality lexically and which were proposed
in the literature review (modal verbs, adverbs, adjectives, nouns, conjunctions and lexical
phrases). This observation is again explainable by the specific genre from which the data
are taken. Most of the sentences are spoken by the narrator. His role is to narrate a
prepared text accurately and clearly in order to elucidate things to laypeople. The simple
structure above serves this goal perfectly. In addition, the subject + verb + that-clause
structure is able to express all evidentiality types, which may also be an explanation for the
high incidence.
In Dutch, by contrast, the simple English syntactic structure is often disturbed in the
subtitles. The subtitles that use volgens or the non-raising use of blijken are translated
according to the same pattern. First, the subject of the original main clause becomes an
adverbial in the subtitle (volgens X, uit/na Y). Second, the subject of the English that-clause
becomes the subject of the Dutch main clause and the evidential verb is deleted in favour of
the verb of the that-clause. This may be seen as the subject that 'raises' from the subject in
the English that-clause to the subject in the Dutch main clause. The term 'raising' is
borrowed from the second structure of blijken, which is also marked by subject to subject
raising.
All these small differences in formulation between the English and the Dutch
sentences are due to the specific evidentiality nature of each language. To express reported
evidence Dutch tends to use the preposition volgens and the raising verb schijnen. In Dutch,
the expression of hearsay evidence is perfectly captured in just one verb: schijnen. Similarly,
the specific verb blijken is used for inference, because its meaning 'to become clear' implies
79
that the speaker has evidence, which causes that something has 'become clear'. The fact
that the structural distinctions are forced by the evidential preferences of Dutch, is proved
by the finding that the English counterpart of volgens, according to, is never used in the
complete corpus. What is more, the default translation of schijnen (seem) is also never
translated with schijnen. Furthermore, albeit these specific words to express Dutch
evidentiality, the word order and the essential message is maintained, as will be emphasised
in the following section.
Next to restructuring the arguments, some subtitlers delete the arguments. The
(partial) deletion of the evidential markers in the subtitles is justified by the subtitle
restrictions, led by the reading speed and synchronicity.
5.3 The functional motivations for evidentiality in popularised
science
The functional motivations proposed in the literature review (acknowledging
responsibility, denying responsibility and assigning authority) were classified as insufficient
when dealing with the actual data. A broad view on evidentiality needs a broad view of
possible motivations to use the evidentiality forms. The engagement types of Martin and
White (2005) (denial, countering, concurrence, pronouncement, endorsement, entertain,
acknowledgement and distancincg) were introduced because it was possible to represent
them all by an evidentiality example of the corpus. Nevertheless, not all engagement types
affect the scientific text equally. When starting from the specific genre, which main
character is the narrator, and its accompanying aim, which is convincing and informing the
unknowing audience, it is noted that the documentary makers must be moved by
endorsement and acknowledgement. These motivations, which are expressed by the
inferential verbs to show, to prove and reporting verbs to say, to believe, explain the high
incidence of the reported evidence type and the inference type.
In spite of the change in type or in structure, the subtitler works according to the
same communicative needs as the speaker. That is why, when the source sentence has to be
condensed, which is frequently the case, the subtitler maintains the most essential
information. In accordance with the aim, this essential information is the statement or the
explanation of a certain scientific finding, which means that the evidential information
becomes accessory. In this view, when dealing with the ten subtitles which delete the
evidential markers completely, one sees that the evidential meaning is not crucial given
80
that the epistemic meaning is not changed. The maintenance of the essential message is led
by the maintenance of the functional motivation. This goal is made perfectly clear by the
exception, which expresses evidentiality in Dutch and not in English. The English corrective
type of denial is used to refute beliefs the audience may have and this is made explicit in the
translation by adding the evidentials men denkt.
Additionally, also the explanation of the vocabulary and the simple syntax in English
as well as in Dutch reinforce the importance of the aim.
81
6 Conclusion
The goal of this dissertation was twofold. In the introduction, two questions were raised:
Does evidentiality occur in English and if it does, does the delineation of the concept change
according to the genre under investigation? The first question was answered in the
literature review. By comparing the different opinions, definitions and taxonomies, it was
decided that English evidentiality was possible but had to be approached in its broadest
sense. This meant that the speaker's epistemic attitude, the evidential must and pragmatic
inference were included in the concept. Secondly, evidentiality was defined in a functional
light. The lexical means to express evidentiality in English were summarised in Figure 2.
This overview was useful as a starting point to compose the corpus and to analyse the data,
taken from the American science documentary How the Earth Was Made. The corpus analysis
consisted of the analysis of the types, the formulations and the functional motivations. The
investigation of these categories applied to the Dutch subtitles served as a meaningful
completion. The results showed that the characteristics outlined in the literature review
were not represented in the corpus. The main outcomes of the corpus analysis and
consequently the main differences with the default definitions and features presented in
the literature review, were explained by two factors: the genre and its aim.
First, the genre is marked by a particular structure on the one hand, and a specific
main character on the other hand. Due to the structure of searching and providing
evidence, the inferential evidence type had to be adjusted to the genre under investigation.
This adjustment was responsible for the dehumanisation of the inferential subjects. On its
turn, these nonhuman subjects are the reason for the exclusion of verbs with a first person
singular. The particular scientific structure was also responsible for the emphasis on the
(un)certainty evidentiality expressions, which proved that these often combine with the
expression of information source. The high incidence of reported and inferential
evidentiality expressions on the other hand, was due to the role of the narrator, who was
moved by the features of endorsement and acknowledgement.
Second, the aim was defined by Meyers (2003: 273) as "convey[ing] scientific
knowledge to a wider audience." This definition implies that the audience consists of
laypeople who want to learn more about a specific scientific subject. The aim explains,
among others, the almost equal division between referring to the specific source (quotative)
and not specifying the source (hearsay), which implies the translation by men, and the
frequent use of the simple structure: subject+ verb+ that-clause.
82
Additionally, the corpus analysis revealed some other important findings. For
instance, some differences between the source and target language could not be explained
by the genre or the special features of evidentials. As a result, it was discovered that small
differences in formulation (blijken, schijnen, volgens) were due to the specific devices used to
express evidentiality in Dutch. The drastic differences (deletion of evidential meaning) were
explained by the subtitle time restrictions. Finally, the study also gained an insight into the
work of the subtitler, who is also affected by the aim and tries to convey the message as
accurately as possible.
In conclusion, evidentiality is possible in English, but it is advisable do adopt a broad
approach and to tailor the analytical toolkit according to the genre under investigation.
This is confirmed by the effect the aim and the genre had on the outcomes of the studies.
Consequently, the evidential types, formulations and the functional motivations will be
different when investigating a different genre. For instance, in spontaneous conversations
the speakers will be more motivated by the 'concurrence' or 'entertain' category. Finally,
the analysis also provided additional information about evidentiality in Dutch, which can be
expressed by schijnen, volgens and blijken. These findings may serve as a starting point for
further investigation about evidentiality in Dutch.
83
7 References
Ahmad, Jameel. 2012. Stylistic features of scientific English: a study of scientific research
articles. English Language & Literature Studies 2(1). 47-55.
Aikhenvald, Alexandra Y. 2004. Evidentiality. Oxford: Oxford University Press.
Aikhenvald, Alexandra Y. & R. M. W. Dixon. 2003. Studies in Evidentiality. Amsterdam:
Benjamins.
Anderson, Lloyd B. 1986. Evidentials, paths of change, and mental maps: typologically
regular asymmetries. In Wallace Chafe & Johanna Nichols (eds.), Evidentiality: The Linguistic
Coding of Epistemology, 273-312, Ablex.
Auwera, Johan van der & Dirk Noël. 2011. Raising: Dutch between English and German.
Journal of Germanic Linguistics 23(1). 1-36.
Baker, R.G. 1982. Monitoring Eye-Movements While Watching Subtitled Television Programs B a
Feasability Study. London: Independent Broadcasting Authority.
Barnes, Janet. 1984. Evidentials in the Tuyuca Verb. International Journal of American
Linguistics 50 (3). 255-271.
Barton, Ellen L. 1993. Evidentials, argumentation, and epistemological stance. College English
55 (7), 745-769.
Boas, Franz, Helene Boas Yampolsky & Zellig S. Harris. 1947. Kwakiutl grammar with a
glossary of the suffixes. Transactions of the American Philosophical Society 37 (3), 203-377.
Chafe, Wallace. 1986. Evidentiality in English conversation and academic writing. In Wallace
Chafe & Johanna Nichols (eds.), Evidentiality: The Linguistic Coding of Epistemology, 261-72,
Ablex.
Chafe, Wallace & Johanna Nichols. 1986. Evidentiality: The Linguistic Coding of Epistemology.
Ablex.
Cornillie, Bert. 2009. Evidentiality and epistemic modality. Functions of language 16 (1), 44 -
62.
84
Dendale, Patrick & Liliane Tasmowski. 2001. Introduction: evidentiality and related notions.
Journal of Pragmatics 33(3), 339-348.
Díaz Cintas, Jorge. 2008. The Didactics of Audiovisual Translation. Amsterdam: Benjamins.
Diewald, Gabriele & Elena Smirnova. 2010. Evidentiality in German: linguistic realization and
regularities in grammaticalization. Berlin: Mouton de Gruyter.
Du Bois, John W. 1986. Self-Evidence and Ritual Speech. In Wallace Chafe & Johanna Nichols
(eds.), Evidentiality: The Linguistic Coding of Epistemology, 313-336. Ablex.
Espasa, Eva. 2004. Myths About Documentary Translation. In Pilar Orero (ed.), Topics in
Audiovisual Translation, 183-197. Amsterdam: Benjamins.
Frățilă, Loredana. 2007. Is scientific discourse purely objective? Romanian Journal of English
Studies 4, 83-88.
Gambier, Yves & Luc van Doorslaer. 2010. Handbook of Translation Studies, Volume 1.
Amsterdam: Benjamins.
Gottlieb, Henrik. 2004. Language-political implications of subtitling. In Pilar Orero (ed.),
Topics in Audiovisual Translation, 83-100. Amsterdam: Benjamins
Haan, Ferdinand de. 2000. Evidentiality in Dutch. In Steve S. Chang, Lily Liaw & Josef
Ruppenhofer (eds.), Proceedings of the Twenty-Fifth Annual Meeting of the Berkeley Linguistics
Society, February 12-15, 1999: General Session and Parasession on Loan Word Phenomena, 74-85.
Berkeley Linguistics Society.
Haan, Ferdinand de. 2001. The Place of inference within the evidential System. International
Journal of American Linguistics 67(2), 193-219.
Ifantidou, Elly. 2001. Evidentials and Relevance. Amsterdam: Benjamins.
Ivarsson, Jan & Mary Carroll. 1998. Subtitling. Simrishamn: Grafo-Tryck AB.
Jacobsen, William H., Jr. 1986. The heterogeneity of evidentials in Makah. In Wallace Chafe
& Johanna Nichols (eds.), Evidentiality: The Linguistic Coding of Epistemology, 3-28. Ablex
Jakobson, Roman. 1966. On Linguistic Aspects of Translation. In Reuben A. Brower (ed.), On
Translation, 232-239. London: Oxford University Press
85
Jakobson, R., L.R. Waugh & M. Halle. 1984. Russian and Slavic Grammar: Studies, 1931-1981.
Mouton Publishers.
Kranich, Svenja. 2011. To hedge or not to hedge: the use of epistemic modal expressions in
popular science in English texts, English-German translations, and German original texts.
Text & Talk: An Interdisciplinary Journal of Language, Discourse & Communication Studies 31(1), 77-
99.
Linde, Zoé de & Neil Kay. 1999. The Semiotics of Subtitling. St. Jerome.
Martin, J.R. & Peter R.R. White. 2005. The Language of Evaluation: Appraisal in English. London:
Palgrave.
Mithun, Marianne. 1986. Evidential diachrony in Northern Iroquoian. In Wallace Chafe &
Johanna Nichols (eds.), Evidentiality: The Linguistic Coding of Epistemology, 89-112. Ablex.
Myers, Greg. 2003. Discourse studies of scientific popularization: questioning the
boundaries. Discourse Studies 5 (2). 265-279.
Orero, Pilar. 2004. Topics in Audiovisual Translation. Amsterdam: Benjamins.
Sanders, José. 2010. Intertwined voices: journalists' modes of representing source
information in journalistic subgenres. English Text Construction 3 (2), 226-249.
Taylor, Christopher. 2004. Multimodal Text Analysis and Subtitling. In Eija Ventola, Cassily
Charles & Martin Kaltenbacher (eds.), Perspectives on Multimodality, 153-72. Amsterdam:
Benjamins.
Van Bogaert, Julie & Patrick Dendale (under review). 2013. Evidential markers in French:
towards a relativistic definition. Studies in Language.
Ventola, Eija, Cassily Charles & Martin Kaltenbacher. 2004. Perspectives on Multimodality.
Amsterdam: Benjamins.
Willett, Thomas. 1988. A cross-linguistic survey of the grammaticization of evidentiality.
Studies in Language: International Journal Sponsored by the Foundation 'Foundations of Language' 12
(1), 51-97.
Wysmantel, Valérie. 2009. Evidentiality in academic versus non-specialist medical and
psychological literature: a comparative study. Lessius PhD thesis.
86